d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
9b9258a12a
freebsd-dev/sys/compat/x86bios/x86bios.c

852 lines
19 KiB
C
Raw Normal View History

Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
/*-
Use a 2 clause BSD-style license instead of stating the code as public domain, as requested by core@ and reviewed by the author.
2009-09-28 08:14:15 +00:00
* Copyright (c) 2009 Alex Keda <admin@lissyara.su>
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
* Copyright (c) 2009-2010 Jung-uk Kim <jkim@FreeBSD.org>
Use a 2 clause BSD-style license instead of stating the code as public domain, as requested by core@ and reviewed by the author.
2009-09-28 08:14:15 +00:00
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_x86bios.h"
#include <sys/param.h>
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
#include <sys/bus.h>
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
#include <sys/kernel.h>
#include <sys/lock.h>
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
#include <sys/malloc.h>
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
#include <sys/module.h>
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
#include <sys/mutex.h>
Add two new debugging tunables for x86bios instead of abusing bootverbose, i.e., debug.x86bios.call and debug.x86bios.int.
2009-12-15 22:44:28 +00:00
#include <sys/sysctl.h>
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
#include <contrib/x86emu/x86emu.h>
#include <contrib/x86emu/x86emu_regs.h>
#include <compat/x86bios/x86bios.h>
#include <dev/pci/pcireg.h>
#include <dev/pci/pcivar.h>
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
#include <vm/vm.h>
#include <vm/pmap.h>
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
#ifdef __amd64__
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
#define X86BIOS_NATIVE_ARCH
#endif
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
#ifdef __i386__
#define X86BIOS_NATIVE_VM86
#endif
#define X86BIOS_MEM_SIZE 0x00100000 /* 1M */
Tidy up printf() calls for debugging.
2010-08-09 22:06:08 +00:00
#define X86BIOS_TRACE(h, n, r) do { \
printf(__STRING(h) \
" (ax=0x%04x bx=0x%04x cx=0x%04x dx=0x%04x es=0x%04x di=0x%04x)\n",\
(n), (r)->R_AX, (r)->R_BX, (r)->R_CX, (r)->R_DX, \
(r)->R_ES, (r)->R_DI); \
} while (0)
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
static struct mtx x86bios_lock;
Mark all SYSCTL_NODEs static that have no corresponding SYSCTL_DECLs. The SYSCTL_NODE macro defines a list that stores all child-elements of that node. If there's no SYSCTL_DECL macro anywhere else, there's no reason why it shouldn't be static.
2011-11-07 15:43:11 +00:00
static SYSCTL_NODE(_debug, OID_AUTO, x86bios, CTLFLAG_RD, NULL,
"x86bios debugging");
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
static int x86bios_trace_call;
Pull in r267961 and r267973 again. Fix for issues reported will follow.
2014-06-28 03:56:17 +00:00
SYSCTL_INT(_debug_x86bios, OID_AUTO, call, CTLFLAG_RWTUN, &x86bios_trace_call, 0,
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
"Trace far function calls");
static int x86bios_trace_int;
Pull in r267961 and r267973 again. Fix for issues reported will follow.
2014-06-28 03:56:17 +00:00
SYSCTL_INT(_debug_x86bios, OID_AUTO, int, CTLFLAG_RWTUN, &x86bios_trace_int, 0,
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
"Trace software interrupt handlers");
#ifdef X86BIOS_NATIVE_VM86
#include <machine/vm86.h>
#include <machine/vmparam.h>
#include <machine/pc/bios.h>
struct vm86context x86bios_vmc;
static void
x86bios_emu2vmf(struct x86emu_regs *regs, struct vm86frame *vmf)
{
vmf->vmf_ds = regs->R_DS;
vmf->vmf_es = regs->R_ES;
vmf->vmf_ax = regs->R_AX;
vmf->vmf_bx = regs->R_BX;
vmf->vmf_cx = regs->R_CX;
vmf->vmf_dx = regs->R_DX;
vmf->vmf_bp = regs->R_BP;
vmf->vmf_si = regs->R_SI;
vmf->vmf_di = regs->R_DI;
}
static void
x86bios_vmf2emu(struct vm86frame *vmf, struct x86emu_regs *regs)
{
regs->R_DS = vmf->vmf_ds;
regs->R_ES = vmf->vmf_es;
Re-add flag register for output. Some BIOS calls actually use it to return success/failure status. Oops.
2010-08-05 19:30:57 +00:00
regs->R_FLG = vmf->vmf_flags;
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
regs->R_AX = vmf->vmf_ax;
regs->R_BX = vmf->vmf_bx;
regs->R_CX = vmf->vmf_cx;
regs->R_DX = vmf->vmf_dx;
regs->R_BP = vmf->vmf_bp;
regs->R_SI = vmf->vmf_si;
regs->R_DI = vmf->vmf_di;
}
void *
x86bios_alloc(uint32_t *offset, size_t size, int flags)
{
Reduce diffs between VM86 and X86EMU wrappers for x86bios_alloc() and x86bios_free(). Add strict sanity checks for VM86 wrapper and add strict page table locking for X86EMU wrapper.
2010-08-09 17:54:26 +00:00
void *vaddr;
x86bios_alloc(): Unsign a counter. The value can't even be signed so we can avoid the signed vs. unsigned comparison. Reviewed by: jkim
2016-04-29 20:22:10 +00:00
u_int i;
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
Reduce diffs between VM86 and X86EMU wrappers for x86bios_alloc() and x86bios_free(). Add strict sanity checks for VM86 wrapper and add strict page table locking for X86EMU wrapper.
2010-08-09 17:54:26 +00:00
if (offset == NULL || size == 0)
return (NULL);
vaddr = contigmalloc(size, M_DEVBUF, flags, 0, X86BIOS_MEM_SIZE,
PAGE_SIZE, 0);
if (vaddr != NULL) {
*offset = vtophys(vaddr);
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
mtx_lock(&x86bios_lock);
Consistently use architecture specific macros.
2010-08-06 15:24:37 +00:00
for (i = 0; i < atop(round_page(size)); i++)
Fix allocation of multiple pages, which forgot to increase page number. Particularly, it caused "vm86_addpage: overlap" panics under VirtualBox. Add a safety check before freeing memory while I am here.
2010-08-06 15:04:01 +00:00
vm86_addpage(&x86bios_vmc, atop(*offset) + i,
Reduce diffs between VM86 and X86EMU wrappers for x86bios_alloc() and x86bios_free(). Add strict sanity checks for VM86 wrapper and add strict page table locking for X86EMU wrapper.
2010-08-09 17:54:26 +00:00
(vm_offset_t)vaddr + ptoa(i));
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
mtx_unlock(&x86bios_lock);
}
Reduce diffs between VM86 and X86EMU wrappers for x86bios_alloc() and x86bios_free(). Add strict sanity checks for VM86 wrapper and add strict page table locking for X86EMU wrapper.
2010-08-09 17:54:26 +00:00
return (vaddr);
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
}
void
x86bios_free(void *addr, size_t size)
{
Reduce diffs between VM86 and X86EMU wrappers for x86bios_alloc() and x86bios_free(). Add strict sanity checks for VM86 wrapper and add strict page table locking for X86EMU wrapper.
2010-08-09 17:54:26 +00:00
vm_paddr_t paddr;
int i, nfree;
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
Reduce diffs between VM86 and X86EMU wrappers for x86bios_alloc() and x86bios_free(). Add strict sanity checks for VM86 wrapper and add strict page table locking for X86EMU wrapper.
2010-08-09 17:54:26 +00:00
if (addr == NULL || size == 0)
return;
paddr = vtophys(addr);
if (paddr >= X86BIOS_MEM_SIZE || (paddr & PAGE_MASK) != 0)
return;
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
mtx_lock(&x86bios_lock);
Reduce diffs between VM86 and X86EMU wrappers for x86bios_alloc() and x86bios_free(). Add strict sanity checks for VM86 wrapper and add strict page table locking for X86EMU wrapper.
2010-08-09 17:54:26 +00:00
for (i = 0; i < x86bios_vmc.npages; i++)
if (x86bios_vmc.pmap[i].kva == (vm_offset_t)addr)
break;
if (i >= x86bios_vmc.npages) {
Fix allocation of multiple pages, which forgot to increase page number. Particularly, it caused "vm86_addpage: overlap" panics under VirtualBox. Add a safety check before freeing memory while I am here.
2010-08-06 15:04:01 +00:00
mtx_unlock(&x86bios_lock);
return;
}
Initialize a variable just before its use.
2010-08-09 18:10:32 +00:00
nfree = atop(round_page(size));
Reduce diffs between VM86 and X86EMU wrappers for x86bios_alloc() and x86bios_free(). Add strict sanity checks for VM86 wrapper and add strict page table locking for X86EMU wrapper.
2010-08-09 17:54:26 +00:00
bzero(x86bios_vmc.pmap + i, sizeof(*x86bios_vmc.pmap) * nfree);
if (i + nfree == x86bios_vmc.npages) {
x86bios_vmc.npages -= nfree;
while (--i >= 0 && x86bios_vmc.pmap[i].kva == 0)
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
x86bios_vmc.npages--;
}
mtx_unlock(&x86bios_lock);
contigfree(addr, size, M_DEVBUF);
}
void
x86bios_init_regs(struct x86regs *regs)
{
bzero(regs, sizeof(*regs));
}
void
x86bios_call(struct x86regs *regs, uint16_t seg, uint16_t off)
{
struct vm86frame vmf;
if (x86bios_trace_call)
Tidy up printf() calls for debugging.
2010-08-09 22:06:08 +00:00
X86BIOS_TRACE(Calling 0x%06x, (seg << 4) + off, regs);
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
bzero(&vmf, sizeof(vmf));
x86bios_emu2vmf((struct x86emu_regs *)regs, &vmf);
vmf.vmf_cs = seg;
vmf.vmf_ip = off;
mtx_lock(&x86bios_lock);
vm86_datacall(-1, &vmf, &x86bios_vmc);
mtx_unlock(&x86bios_lock);
x86bios_vmf2emu(&vmf, (struct x86emu_regs *)regs);
if (x86bios_trace_call)
Tidy up printf() calls for debugging.
2010-08-09 22:06:08 +00:00
X86BIOS_TRACE(Exiting 0x%06x, (seg << 4) + off, regs);
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
}
uint32_t
x86bios_get_intr(int intno)
{
Optimize interrupt vector lookup. There is no need to check the page table.
2010-08-07 03:45:45 +00:00
return (readl(BIOS_PADDRTOVADDR(intno * 4)));
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
}
Add x86bios_set_intr() to set interrupt vectors for real mode and simplify x86bios_get_intr() a little.
2010-08-25 21:03:50 +00:00
void
x86bios_set_intr(int intno, uint32_t saddr)
{
writel(BIOS_PADDRTOVADDR(intno * 4), saddr);
}
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
void
x86bios_intr(struct x86regs *regs, int intno)
{
struct vm86frame vmf;
if (x86bios_trace_int)
Tidy up printf() calls for debugging.
2010-08-09 22:06:08 +00:00
X86BIOS_TRACE(Calling INT 0x%02x, intno, regs);
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
bzero(&vmf, sizeof(vmf));
x86bios_emu2vmf((struct x86emu_regs *)regs, &vmf);
mtx_lock(&x86bios_lock);
vm86_datacall(intno, &vmf, &x86bios_vmc);
mtx_unlock(&x86bios_lock);
x86bios_vmf2emu(&vmf, (struct x86emu_regs *)regs);
if (x86bios_trace_int)
Tidy up printf() calls for debugging.
2010-08-09 22:06:08 +00:00
X86BIOS_TRACE(Exiting INT 0x%02x, intno, regs);
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
}
void *
x86bios_offset(uint32_t offset)
{
vm_offset_t addr;
addr = vm86_getaddr(&x86bios_vmc, X86BIOS_PHYSTOSEG(offset),
X86BIOS_PHYSTOOFF(offset));
if (addr == 0)
addr = BIOS_PADDRTOVADDR(offset);
return ((void *)addr);
}
static int
x86bios_init(void)
{
mtx_init(&x86bios_lock, "x86bios lock", NULL, MTX_DEF);
bzero(&x86bios_vmc, sizeof(x86bios_vmc));
return (0);
}
static int
x86bios_uninit(void)
{
mtx_destroy(&x86bios_lock);
return (0);
}
#else
#include <machine/iodev.h>
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
#define X86BIOS_PAGE_SIZE 0x00001000 /* 4K */
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
#define X86BIOS_IVT_SIZE 0x00000500 /* 1K + 256 (BDA) */
#define X86BIOS_IVT_BASE 0x00000000
#define X86BIOS_RAM_BASE 0x00001000
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
#define X86BIOS_ROM_BASE 0x000a0000
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
Tidy up printf() calls for debugging.
2010-08-09 22:06:08 +00:00
#define X86BIOS_ROM_SIZE (X86BIOS_MEM_SIZE - x86bios_rom_phys)
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
#define X86BIOS_SEG_SIZE X86BIOS_PAGE_SIZE
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
#define X86BIOS_PAGES (X86BIOS_MEM_SIZE / X86BIOS_PAGE_SIZE)
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
#define X86BIOS_R_SS _pad2
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
#define X86BIOS_R_SP _pad3.I16_reg.x_reg
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
static struct x86emu x86bios_emu;
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
static void *x86bios_ivt;
static void *x86bios_rom;
static void *x86bios_seg;
static vm_offset_t *x86bios_map;
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
static vm_paddr_t x86bios_rom_phys;
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
static vm_paddr_t x86bios_seg_phys;
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
static int x86bios_fault;
static uint32_t x86bios_fault_addr;
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
static uint16_t x86bios_fault_cs;
static uint16_t x86bios_fault_ip;
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
static void
x86bios_set_fault(struct x86emu *emu, uint32_t addr)
{
x86bios_fault = 1;
x86bios_fault_addr = addr;
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
x86bios_fault_cs = emu->x86.R_CS;
x86bios_fault_ip = emu->x86.R_IP;
x86emu_halt_sys(emu);
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
}
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
static void *
x86bios_get_pages(uint32_t offset, size_t size)
{
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
vm_offset_t addr;
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
Support memory wraparound instead of high memory as VM86 mode does. Suggested by: delphij
2010-03-22 18:43:36 +00:00
if (offset + size > X86BIOS_MEM_SIZE + X86BIOS_IVT_SIZE)
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
return (NULL);
Support memory wraparound instead of high memory as VM86 mode does. Suggested by: delphij
2010-03-22 18:43:36 +00:00
if (offset >= X86BIOS_MEM_SIZE)
offset -= X86BIOS_MEM_SIZE;
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
addr = x86bios_map[offset / X86BIOS_PAGE_SIZE];
if (addr != 0)
addr += offset % X86BIOS_PAGE_SIZE;
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
return ((void *)addr);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
}
static void
x86bios_set_pages(vm_offset_t va, vm_paddr_t pa, size_t size)
{
int i, j;
for (i = pa / X86BIOS_PAGE_SIZE, j = 0;
j < howmany(size, X86BIOS_PAGE_SIZE); i++, j++)
x86bios_map[i] = va + j * X86BIOS_PAGE_SIZE;
}
static uint8_t
x86bios_emu_rdb(struct x86emu *emu, uint32_t addr)
{
uint8_t *va;
va = x86bios_get_pages(addr, sizeof(*va));
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
if (va == NULL)
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
x86bios_set_fault(emu, addr);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
return (*va);
}
static uint16_t
x86bios_emu_rdw(struct x86emu *emu, uint32_t addr)
{
uint16_t *va;
va = x86bios_get_pages(addr, sizeof(*va));
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
if (va == NULL)
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
x86bios_set_fault(emu, addr);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
#ifndef __NO_STRICT_ALIGNMENT
if ((addr & 1) != 0)
return (le16dec(va));
else
#endif
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
return (le16toh(*va));
}
static uint32_t
x86bios_emu_rdl(struct x86emu *emu, uint32_t addr)
{
uint32_t *va;
va = x86bios_get_pages(addr, sizeof(*va));
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
if (va == NULL)
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
x86bios_set_fault(emu, addr);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
#ifndef __NO_STRICT_ALIGNMENT
if ((addr & 3) != 0)
return (le32dec(va));
else
#endif
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
return (le32toh(*va));
}
static void
x86bios_emu_wrb(struct x86emu *emu, uint32_t addr, uint8_t val)
{
uint8_t *va;
va = x86bios_get_pages(addr, sizeof(*va));
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
if (va == NULL)
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
x86bios_set_fault(emu, addr);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
*va = val;
}
static void
x86bios_emu_wrw(struct x86emu *emu, uint32_t addr, uint16_t val)
{
uint16_t *va;
va = x86bios_get_pages(addr, sizeof(*va));
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
if (va == NULL)
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
x86bios_set_fault(emu, addr);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
#ifndef __NO_STRICT_ALIGNMENT
if ((addr & 1) != 0)
le16enc(va, val);
else
#endif
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
*va = htole16(val);
}
static void
x86bios_emu_wrl(struct x86emu *emu, uint32_t addr, uint32_t val)
{
uint32_t *va;
va = x86bios_get_pages(addr, sizeof(*va));
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
if (va == NULL)
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
x86bios_set_fault(emu, addr);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
#ifndef __NO_STRICT_ALIGNMENT
if ((addr & 3) != 0)
le32enc(va, val);
else
#endif
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
*va = htole32(val);
}
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
static uint8_t
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
x86bios_emu_inb(struct x86emu *emu, uint16_t port)
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
{
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
Do not block any I/O port on amd64.
2010-08-07 04:05:58 +00:00
#ifndef X86BIOS_NATIVE_ARCH
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
if (port == 0xb2) /* APM scratch register */
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
return (0);
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
if (port >= 0x80 && port < 0x88) /* POST status register */
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
return (0);
Do not block any I/O port on amd64.
2010-08-07 04:05:58 +00:00
#endif
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
return (iodev_read_1(port));
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
}
static uint16_t
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
x86bios_emu_inw(struct x86emu *emu, uint16_t port)
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
{
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
uint16_t val;
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
Do not block any I/O port on amd64.
2010-08-07 04:05:58 +00:00
#ifndef X86BIOS_NATIVE_ARCH
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
if (port >= 0x80 && port < 0x88) /* POST status register */
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
return (0);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
if ((port & 1) != 0) {
val = iodev_read_1(port);
val |= iodev_read_1(port + 1) << 8;
} else
#endif
val = iodev_read_2(port);
return (val);
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
}
static uint32_t
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
x86bios_emu_inl(struct x86emu *emu, uint16_t port)
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
{
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
uint32_t val;
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
Do not block any I/O port on amd64.
2010-08-07 04:05:58 +00:00
#ifndef X86BIOS_NATIVE_ARCH
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
if (port >= 0x80 && port < 0x88) /* POST status register */
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
return (0);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
if ((port & 1) != 0) {
val = iodev_read_1(port);
val |= iodev_read_2(port + 1) << 8;
val |= iodev_read_1(port + 3) << 24;
} else if ((port & 2) != 0) {
val = iodev_read_2(port);
val |= iodev_read_2(port + 2) << 16;
} else
#endif
val = iodev_read_4(port);
return (val);
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
}
static void
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
x86bios_emu_outb(struct x86emu *emu, uint16_t port, uint8_t val)
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
{
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
Do not block any I/O port on amd64.
2010-08-07 04:05:58 +00:00
#ifndef X86BIOS_NATIVE_ARCH
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
if (port == 0xb2) /* APM scratch register */
return;
if (port >= 0x80 && port < 0x88) /* POST status register */
return;
Do not block any I/O port on amd64.
2010-08-07 04:05:58 +00:00
#endif
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
iodev_write_1(port, val);
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
}
static void
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
x86bios_emu_outw(struct x86emu *emu, uint16_t port, uint16_t val)
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
{
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
Do not block any I/O port on amd64.
2010-08-07 04:05:58 +00:00
#ifndef X86BIOS_NATIVE_ARCH
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
if (port >= 0x80 && port < 0x88) /* POST status register */
return;
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
if ((port & 1) != 0) {
iodev_write_1(port, val);
iodev_write_1(port + 1, val >> 8);
} else
#endif
iodev_write_2(port, val);
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
}
static void
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
x86bios_emu_outl(struct x86emu *emu, uint16_t port, uint32_t val)
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
{
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
Do not block any I/O port on amd64.
2010-08-07 04:05:58 +00:00
#ifndef X86BIOS_NATIVE_ARCH
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
if (port >= 0x80 && port < 0x88) /* POST status register */
return;
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
if ((port & 1) != 0) {
iodev_write_1(port, val);
iodev_write_2(port + 1, val >> 8);
iodev_write_1(port + 3, val >> 24);
} else if ((port & 2) != 0) {
iodev_write_2(port, val);
iodev_write_2(port + 2, val >> 16);
} else
#endif
iodev_write_4(port, val);
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
}
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
void *
Let x86bios_alloc() pass contigmalloc(9) flags. Use it to set M_WAITOK from VESA BIOS initialization. All other malloc(9) uses in the function is blocking any way.
2010-06-23 17:20:51 +00:00
x86bios_alloc(uint32_t *offset, size_t size, int flags)
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
{
void *vaddr;
if (offset == NULL || size == 0)
return (NULL);
Let x86bios_alloc() pass contigmalloc(9) flags. Use it to set M_WAITOK from VESA BIOS initialization. All other malloc(9) uses in the function is blocking any way.
2010-06-23 17:20:51 +00:00
vaddr = contigmalloc(size, M_DEVBUF, flags, X86BIOS_RAM_BASE,
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
x86bios_rom_phys, X86BIOS_PAGE_SIZE, 0);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
if (vaddr != NULL) {
*offset = vtophys(vaddr);
Tidy up locking and memory allocation for the real mode emulator wrapper. Now we use a regular mutex instead of a spin mutex. When we enter and exit the emulator, spinlock_enter() and spinlock_exit() are additionally used. Move some page table related stuff from x86bios_init() and x86bios_uninit() to x86bios_map_mem() and x86bios_unmap_mem().
2010-08-10 06:25:08 +00:00
mtx_lock(&x86bios_lock);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
x86bios_set_pages((vm_offset_t)vaddr, *offset, size);
Tidy up locking and memory allocation for the real mode emulator wrapper. Now we use a regular mutex instead of a spin mutex. When we enter and exit the emulator, spinlock_enter() and spinlock_exit() are additionally used. Move some page table related stuff from x86bios_init() and x86bios_uninit() to x86bios_map_mem() and x86bios_unmap_mem().
2010-08-10 06:25:08 +00:00
mtx_unlock(&x86bios_lock);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
}
return (vaddr);
}
void
x86bios_free(void *addr, size_t size)
{
vm_paddr_t paddr;
if (addr == NULL || size == 0)
return;
paddr = vtophys(addr);
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
if (paddr < X86BIOS_RAM_BASE || paddr >= x86bios_rom_phys ||
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
paddr % X86BIOS_PAGE_SIZE != 0)
return;
Tidy up locking and memory allocation for the real mode emulator wrapper. Now we use a regular mutex instead of a spin mutex. When we enter and exit the emulator, spinlock_enter() and spinlock_exit() are additionally used. Move some page table related stuff from x86bios_init() and x86bios_uninit() to x86bios_map_mem() and x86bios_unmap_mem().
2010-08-10 06:25:08 +00:00
mtx_lock(&x86bios_lock);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
bzero(x86bios_map + paddr / X86BIOS_PAGE_SIZE,
sizeof(*x86bios_map) * howmany(size, X86BIOS_PAGE_SIZE));
Tidy up locking and memory allocation for the real mode emulator wrapper. Now we use a regular mutex instead of a spin mutex. When we enter and exit the emulator, spinlock_enter() and spinlock_exit() are additionally used. Move some page table related stuff from x86bios_init() and x86bios_uninit() to x86bios_map_mem() and x86bios_unmap_mem().
2010-08-10 06:25:08 +00:00
mtx_unlock(&x86bios_lock);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
contigfree(addr, size, M_DEVBUF);
}
void
x86bios_init_regs(struct x86regs *regs)
{
bzero(regs, sizeof(*regs));
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
regs->X86BIOS_R_SS = X86BIOS_PHYSTOSEG(x86bios_seg_phys);
regs->X86BIOS_R_SP = X86BIOS_PAGE_SIZE - 2;
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
}
void
x86bios_call(struct x86regs *regs, uint16_t seg, uint16_t off)
{
Add two new debugging tunables for x86bios instead of abusing bootverbose, i.e., debug.x86bios.call and debug.x86bios.int.
2009-12-15 22:44:28 +00:00
if (x86bios_trace_call)
Tidy up printf() calls for debugging.
2010-08-09 22:06:08 +00:00
X86BIOS_TRACE(Calling 0x%06x, (seg << 4) + off, regs);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
Tidy up locking and memory allocation for the real mode emulator wrapper. Now we use a regular mutex instead of a spin mutex. When we enter and exit the emulator, spinlock_enter() and spinlock_exit() are additionally used. Move some page table related stuff from x86bios_init() and x86bios_uninit() to x86bios_map_mem() and x86bios_unmap_mem().
2010-08-10 06:25:08 +00:00
mtx_lock(&x86bios_lock);
Silence VPS-Studio errors (V512). These buffer underflows are intentional.
2016-02-18 19:37:39 +00:00
memcpy((struct x86regs *)&x86bios_emu.x86, regs, sizeof(*regs));
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
x86bios_fault = 0;
Place spinlock_enter() and spinlock_exit() just around X86EMU calls.
2010-08-10 15:22:48 +00:00
spinlock_enter();
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
x86emu_exec_call(&x86bios_emu, seg, off);
Tidy up locking and memory allocation for the real mode emulator wrapper. Now we use a regular mutex instead of a spin mutex. When we enter and exit the emulator, spinlock_enter() and spinlock_exit() are additionally used. Move some page table related stuff from x86bios_init() and x86bios_uninit() to x86bios_map_mem() and x86bios_unmap_mem().
2010-08-10 06:25:08 +00:00
spinlock_exit();
Place spinlock_enter() and spinlock_exit() just around X86EMU calls.
2010-08-10 15:22:48 +00:00
memcpy(regs, &x86bios_emu.x86, sizeof(*regs));
Tidy up locking and memory allocation for the real mode emulator wrapper. Now we use a regular mutex instead of a spin mutex. When we enter and exit the emulator, spinlock_enter() and spinlock_exit() are additionally used. Move some page table related stuff from x86bios_init() and x86bios_uninit() to x86bios_map_mem() and x86bios_unmap_mem().
2010-08-10 06:25:08 +00:00
mtx_unlock(&x86bios_lock);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
if (x86bios_trace_call) {
Tidy up printf() calls for debugging.
2010-08-09 22:06:08 +00:00
X86BIOS_TRACE(Exiting 0x%06x, (seg << 4) + off, regs);
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
if (x86bios_fault)
Tidy up printf() calls for debugging.
2010-08-09 22:06:08 +00:00
printf("Page fault at 0x%06x from 0x%04x:0x%04x.\n",
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
x86bios_fault_addr, x86bios_fault_cs,
x86bios_fault_ip);
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
}
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
}
uint32_t
x86bios_get_intr(int intno)
{
Add x86bios_set_intr() to set interrupt vectors for real mode and simplify x86bios_get_intr() a little.
2010-08-25 21:03:50 +00:00
return (le32toh(*((uint32_t *)x86bios_ivt + intno)));
}
void
x86bios_set_intr(int intno, uint32_t saddr)
{
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
Add x86bios_set_intr() to set interrupt vectors for real mode and simplify x86bios_get_intr() a little.
2010-08-25 21:03:50 +00:00
*((uint32_t *)x86bios_ivt + intno) = htole32(saddr);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
}
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
void
- Use FreeBSD function naming convention. - Change x86biosCall() to more appropriate x86bios_intr().[1] Discussed with: delphij, paradox (ddkprog yahoo com) Submitted by: paradox (ddkprog yahoo com)[1]
2009-09-24 19:24:42 +00:00
x86bios_intr(struct x86regs *regs, int intno)
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
{
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
if (intno < 0 || intno > 255)
return;
Add two new debugging tunables for x86bios instead of abusing bootverbose, i.e., debug.x86bios.call and debug.x86bios.int.
2009-12-15 22:44:28 +00:00
if (x86bios_trace_int)
Tidy up printf() calls for debugging.
2010-08-09 22:06:08 +00:00
X86BIOS_TRACE(Calling INT 0x%02x, intno, regs);
Dump real mode registers under bootverbose to help debugging BIOS emulator.
2009-09-24 22:42:35 +00:00
Tidy up locking and memory allocation for the real mode emulator wrapper. Now we use a regular mutex instead of a spin mutex. When we enter and exit the emulator, spinlock_enter() and spinlock_exit() are additionally used. Move some page table related stuff from x86bios_init() and x86bios_uninit() to x86bios_map_mem() and x86bios_unmap_mem().
2010-08-10 06:25:08 +00:00
mtx_lock(&x86bios_lock);
Silence VPS-Studio errors (V512). These buffer underflows are intentional.
2016-02-18 19:37:39 +00:00
memcpy((struct x86regs *)&x86bios_emu.x86, regs, sizeof(*regs));
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
x86bios_fault = 0;
Place spinlock_enter() and spinlock_exit() just around X86EMU calls.
2010-08-10 15:22:48 +00:00
spinlock_enter();
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
x86emu_exec_intr(&x86bios_emu, intno);
Tidy up locking and memory allocation for the real mode emulator wrapper. Now we use a regular mutex instead of a spin mutex. When we enter and exit the emulator, spinlock_enter() and spinlock_exit() are additionally used. Move some page table related stuff from x86bios_init() and x86bios_uninit() to x86bios_map_mem() and x86bios_unmap_mem().
2010-08-10 06:25:08 +00:00
spinlock_exit();
Place spinlock_enter() and spinlock_exit() just around X86EMU calls.
2010-08-10 15:22:48 +00:00
memcpy(regs, &x86bios_emu.x86, sizeof(*regs));
Tidy up locking and memory allocation for the real mode emulator wrapper. Now we use a regular mutex instead of a spin mutex. When we enter and exit the emulator, spinlock_enter() and spinlock_exit() are additionally used. Move some page table related stuff from x86bios_init() and x86bios_uninit() to x86bios_map_mem() and x86bios_unmap_mem().
2010-08-10 06:25:08 +00:00
mtx_unlock(&x86bios_lock);
Dump real mode registers under bootverbose to help debugging BIOS emulator.
2009-09-24 22:42:35 +00:00
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
if (x86bios_trace_int) {
Tidy up printf() calls for debugging.
2010-08-09 22:06:08 +00:00
X86BIOS_TRACE(Exiting INT 0x%02x, intno, regs);
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
if (x86bios_fault)
Tidy up printf() calls for debugging.
2010-08-09 22:06:08 +00:00
printf("Page fault at 0x%06x from 0x%04x:0x%04x.\n",
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
x86bios_fault_addr, x86bios_fault_cs,
x86bios_fault_ip);
Detect illegal access to unmapped memory within real mode emulator to aid debugging. Update copyright date while I am here.
2010-03-18 20:15:34 +00:00
}
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
}
void *
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
x86bios_offset(uint32_t offset)
{
return (x86bios_get_pages(offset, 1));
}
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
static __inline void
x86bios_unmap_mem(void)
{
Tidy up locking and memory allocation for the real mode emulator wrapper. Now we use a regular mutex instead of a spin mutex. When we enter and exit the emulator, spinlock_enter() and spinlock_exit() are additionally used. Move some page table related stuff from x86bios_init() and x86bios_uninit() to x86bios_map_mem() and x86bios_unmap_mem().
2010-08-10 06:25:08 +00:00
free(x86bios_map, M_DEVBUF);
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
if (x86bios_ivt != NULL)
#ifdef X86BIOS_NATIVE_ARCH
Prefer pmap_unmapbios() over pmap_unmapdev(). The binary does not change after this because pmap_unmapbios() is a macro for pmap_unmapdev() on amd64.
2010-10-05 18:38:23 +00:00
pmap_unmapbios((vm_offset_t)x86bios_ivt, X86BIOS_IVT_SIZE);
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
#else
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
free(x86bios_ivt, M_DEVBUF);
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
#endif
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
if (x86bios_rom != NULL)
pmap_unmapdev((vm_offset_t)x86bios_rom, X86BIOS_ROM_SIZE);
if (x86bios_seg != NULL)
contigfree(x86bios_seg, X86BIOS_SEG_SIZE, M_DEVBUF);
}
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
static __inline int
x86bios_map_mem(void)
{
Tidy up locking and memory allocation for the real mode emulator wrapper. Now we use a regular mutex instead of a spin mutex. When we enter and exit the emulator, spinlock_enter() and spinlock_exit() are additionally used. Move some page table related stuff from x86bios_init() and x86bios_uninit() to x86bios_map_mem() and x86bios_unmap_mem().
2010-08-10 06:25:08 +00:00
x86bios_map = malloc(sizeof(*x86bios_map) * X86BIOS_PAGES, M_DEVBUF,
M_WAITOK | M_ZERO);
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
#ifdef X86BIOS_NATIVE_ARCH
Fix a copy-and-pasto in the previous commit.
2009-10-19 21:01:42 +00:00
x86bios_ivt = pmap_mapbios(X86BIOS_IVT_BASE, X86BIOS_IVT_SIZE);
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
/* Probe EBDA via BDA. */
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
x86bios_rom_phys = *(uint16_t *)((caddr_t)x86bios_ivt + 0x40e);
x86bios_rom_phys = x86bios_rom_phys << 4;
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
if (x86bios_rom_phys != 0 && x86bios_rom_phys < X86BIOS_ROM_BASE &&
X86BIOS_ROM_BASE - x86bios_rom_phys <= 128 * 1024)
x86bios_rom_phys =
rounddown(x86bios_rom_phys, X86BIOS_PAGE_SIZE);
else
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
#else
x86bios_ivt = malloc(X86BIOS_IVT_SIZE, M_DEVBUF, M_ZERO | M_WAITOK);
#endif
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
x86bios_rom_phys = X86BIOS_ROM_BASE;
x86bios_rom = pmap_mapdev(x86bios_rom_phys, X86BIOS_ROM_SIZE);
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
if (x86bios_rom == NULL)
goto fail;
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
#ifdef X86BIOS_NATIVE_ARCH
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
/* Change attribute for EBDA. */
if (x86bios_rom_phys < X86BIOS_ROM_BASE &&
pmap_change_attr((vm_offset_t)x86bios_rom,
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
X86BIOS_ROM_BASE - x86bios_rom_phys, PAT_WRITE_BACK) != 0)
goto fail;
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
#endif
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
x86bios_seg = contigmalloc(X86BIOS_SEG_SIZE, M_DEVBUF, M_WAITOK,
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
X86BIOS_RAM_BASE, x86bios_rom_phys, X86BIOS_PAGE_SIZE, 0);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
x86bios_seg_phys = vtophys(x86bios_seg);
Tidy up locking and memory allocation for the real mode emulator wrapper. Now we use a regular mutex instead of a spin mutex. When we enter and exit the emulator, spinlock_enter() and spinlock_exit() are additionally used. Move some page table related stuff from x86bios_init() and x86bios_uninit() to x86bios_map_mem() and x86bios_unmap_mem().
2010-08-10 06:25:08 +00:00
x86bios_set_pages((vm_offset_t)x86bios_ivt, X86BIOS_IVT_BASE,
X86BIOS_IVT_SIZE);
x86bios_set_pages((vm_offset_t)x86bios_rom, x86bios_rom_phys,
X86BIOS_ROM_SIZE);
x86bios_set_pages((vm_offset_t)x86bios_seg, x86bios_seg_phys,
X86BIOS_SEG_SIZE);
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
if (bootverbose) {
Tidy up printf() calls for debugging.
2010-08-09 22:06:08 +00:00
printf("x86bios: IVT 0x%06jx-0x%06jx at %p\n",
(vm_paddr_t)X86BIOS_IVT_BASE,
(vm_paddr_t)X86BIOS_IVT_SIZE + X86BIOS_IVT_BASE - 1,
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
x86bios_ivt);
Tidy up printf() calls for debugging.
2010-08-09 22:06:08 +00:00
printf("x86bios: SSEG 0x%06jx-0x%06jx at %p\n",
x86bios_seg_phys,
(vm_paddr_t)X86BIOS_SEG_SIZE + x86bios_seg_phys - 1,
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
x86bios_seg);
if (x86bios_rom_phys < X86BIOS_ROM_BASE)
Tidy up printf() calls for debugging.
2010-08-09 22:06:08 +00:00
printf("x86bios: EBDA 0x%06jx-0x%06jx at %p\n",
x86bios_rom_phys, (vm_paddr_t)X86BIOS_ROM_BASE - 1,
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
x86bios_rom);
Tidy up printf() calls for debugging.
2010-08-09 22:06:08 +00:00
printf("x86bios: ROM 0x%06jx-0x%06jx at %p\n",
(vm_paddr_t)X86BIOS_ROM_BASE,
(vm_paddr_t)X86BIOS_MEM_SIZE - X86BIOS_SEG_SIZE - 1,
(caddr_t)x86bios_rom + X86BIOS_ROM_BASE - x86bios_rom_phys);
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
}
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
return (0);
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
fail:
x86bios_unmap_mem();
- Map EBDA if available and add 64KB above 1MB (high memory), just in case. - Print the initial memory map when bootverbose is set. - Change the page fault address format from linear to %cs:%ip style. - Move duplicate code into a newly added function. - Add strictly aligned memory access for distant future. ;-)
2010-03-19 21:15:43 +00:00
Reduce MD code further. At least, it compiles on ia64 now (but it is not connected to build). The idea/code was shamelessly taken from r207329.
2010-05-01 01:05:07 +00:00
return (1);
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
}
Do not initialize mutex and return error if it cannot map memory.
2010-05-01 00:36:40 +00:00
static int
x86bios_init(void)
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
{
Tidy up locking and memory allocation for the real mode emulator wrapper. Now we use a regular mutex instead of a spin mutex. When we enter and exit the emulator, spinlock_enter() and spinlock_exit() are additionally used. Move some page table related stuff from x86bios_init() and x86bios_uninit() to x86bios_map_mem() and x86bios_unmap_mem().
2010-08-10 06:25:08 +00:00
mtx_init(&x86bios_lock, "x86bios lock", NULL, MTX_DEF);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
if (x86bios_map_mem() != 0)
Do not initialize mutex and return error if it cannot map memory.
2010-05-01 00:36:40 +00:00
return (ENOMEM);
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
bzero(&x86bios_emu, sizeof(x86bios_emu));
x86bios_emu.emu_rdb = x86bios_emu_rdb;
x86bios_emu.emu_rdw = x86bios_emu_rdw;
x86bios_emu.emu_rdl = x86bios_emu_rdl;
x86bios_emu.emu_wrb = x86bios_emu_wrb;
x86bios_emu.emu_wrw = x86bios_emu_wrw;
x86bios_emu.emu_wrl = x86bios_emu_wrl;
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
- Use spin lock instead of default mutex for safety. INT/IRET instructions save/clear/restore flags but emulated flags have no effect on the host. I believe BIOS writers never meant to run their code in emulated environment with interrupt enabled. :-) - Use memcpy(3) instead of copying individual members. I believe struct x86regs was intentionally copied from the first half of struct x86emu_regs for this very purpose. - Fix some style nits and consistencies. Reviewed by: delphij, paradox (ddkprog yahoo com)
2009-09-23 20:13:36 +00:00
x86bios_emu.emu_inb = x86bios_emu_inb;
x86bios_emu.emu_inw = x86bios_emu_inw;
x86bios_emu.emu_inl = x86bios_emu_inl;
x86bios_emu.emu_outb = x86bios_emu_outb;
x86bios_emu.emu_outw = x86bios_emu_outw;
x86bios_emu.emu_outl = x86bios_emu_outl;
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
Do not initialize mutex and return error if it cannot map memory.
2010-05-01 00:36:40 +00:00
return (0);
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
}
Do not initialize mutex and return error if it cannot map memory.
2010-05-01 00:36:40 +00:00
static int
x86bios_uninit(void)
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
{
Tidy up locking and memory allocation for the real mode emulator wrapper. Now we use a regular mutex instead of a spin mutex. When we enter and exit the emulator, spinlock_enter() and spinlock_exit() are additionally used. Move some page table related stuff from x86bios_init() and x86bios_uninit() to x86bios_map_mem() and x86bios_unmap_mem().
2010-08-10 06:25:08 +00:00
x86bios_unmap_mem();
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
mtx_destroy(&x86bios_lock);
Do not initialize mutex and return error if it cannot map memory.
2010-05-01 00:36:40 +00:00
return (0);
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
}
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
#endif
void *
x86bios_get_orm(uint32_t offset)
{
uint8_t *p;
/* Does the shadow ROM contain BIOS POST code for x86? */
p = x86bios_offset(offset);
Check opcode for short jump as well. Some option ROMs do short jumps (e.g., some NVIDIA video cards) and we were not able to do POST while resuming because we only honored long jump. MFC after: 3 days
2010-08-25 20:52:40 +00:00
if (p == NULL || p[0] != 0x55 || p[1] != 0xaa ||
(p[3] != 0xe9 && p[3] != 0xeb))
Implement a simple native VM86 backend for X86BIOS. Now i386 uses native VM86 calls instead of the real mode emulator as a backend. VM86 has been proven reliable for very long time and it is actually few times faster than emulation. Increase maximum number of page table entries per VM86 context from 3 to 8 pages. It was (ridiculously) low and insufficient for new VM86 backend, which shares one context globally. Slighly rearrange and clean up the emulator backend to accommodate new code. The only visible change here is stack size, which is decreased from 64K to 4K bytes to sync. with VM86. Actually, it seems there is no need for big stack in real mode. MFC after: 1 month
2010-08-05 18:48:30 +00:00
return (NULL);
return (p);
}
int
x86bios_match_device(uint32_t offset, device_t dev)
{
uint8_t *p;
uint16_t device, vendor;
uint8_t class, progif, subclass;
/* Does the shadow ROM contain BIOS POST code for x86? */
p = x86bios_get_orm(offset);
if (p == NULL)
return (0);
/* Does it contain PCI data structure? */
p += le16toh(*(uint16_t *)(p + 0x18));
if (bcmp(p, "PCIR", 4) != 0 ||
le16toh(*(uint16_t *)(p + 0x0a)) < 0x18 || *(p + 0x14) != 0)
return (0);
/* Does it match the vendor, device, and classcode? */
vendor = le16toh(*(uint16_t *)(p + 0x04));
device = le16toh(*(uint16_t *)(p + 0x06));
progif = *(p + 0x0d);
subclass = *(p + 0x0e);
class = *(p + 0x0f);
if (vendor != pci_get_vendor(dev) || device != pci_get_device(dev) ||
class != pci_get_class(dev) || subclass != pci_get_subclass(dev) ||
progif != pci_get_progif(dev))
return (0);
return (1);
}
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
static int
x86bios_modevent(module_t mod __unused, int type, void *data __unused)
{
switch (type) {
case MOD_LOAD:
Do not initialize mutex and return error if it cannot map memory.
2010-05-01 00:36:40 +00:00
return (x86bios_init());
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
case MOD_UNLOAD:
Do not initialize mutex and return error if it cannot map memory.
2010-05-01 00:36:40 +00:00
return (x86bios_uninit());
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
default:
Rewrite x86bios and update its dependent drivers. - Do not map entire real mode memory (1MB). Instead, we map IVT/BDA and ROM area separately. Most notably, ROM area is mapped as device memory (uncacheable) as it should be. User memory is dynamically allocated and free'ed with contigmalloc(9) and contigfree(9). Remove now redundant and potentially dangerous x86bios_alloc.c. If this emulator ever grows to support non-PC hardware, we may implement it with rman(9) later. - Move all host-specific initializations from x86emu_util.c to x86bios.c and remove now unnecessary x86emu_util.c. Currently, non-PC hardware is not supported. We may use bus_space(9) later when the KPI is fixed. - Replace all bzero() calls for emulated registers with more obviously named x86bios_init_regs(). This function also initializes DS and SS properly. - Add x86bios_get_intr(). This function checks if the interrupt vector is available for the platform. It is not necessary for PC-compatible hardware but it may be needed later. ;-) - Do not try turning off monitor if DPMS does not support the state. - Allocate stable memory for VESA OEM strings instead of just holding pointers to them. They may or may not be accessible always. Fix a memory leak of video mode table while I am here. - Add (experimental) BIOS POST call for vesa(4). This function calls VGA BIOS POST code from the current VGA option ROM. Some video controllers cannot save and restore the state properly even if it is claimed to be supported. Usually the symptom is blank display after resuming from suspend state. If the video mode does not match the previous mode after restoring, we try BIOS POST and force the known good initial state. Some magic was taken from NetBSD (and it was taken from vbetool, I believe.) - Add a loader tunable for vgapci(4) to give a hint to dpms(4) and vesa(4) to identify who owns the VESA BIOS. This is very useful for multi-display adapter setup. By default, the POST video controller is automatically probed and the tunable "hw.pci.default_vgapci_unit" is set to corresponding vgapci unit number. You may override it from loader but it is very unlikely to be necessary. Unfortunately only AGP/PCI/PCI-E controllers can be matched because ISA controller does not have necessary device IDs. - Fix a long standing bug in state save/restore function. The state buffer pointer should be ES:BX, not ES:DI according to VBE 3.0. If it ever worked, that's because BX was always zero. :-) - Clean up register initializations more clearer per VBE 3.0. - Fix a lot of style issues with vesa(4).
2009-10-19 20:58:10 +00:00
return (ENOTSUP);
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
}
}
static moduledata_t x86bios_mod = {
"x86bios",
x86bios_modevent,
NULL,
};
To be safe, make this module initialize earlier by specifying SI_SUB_CPU instead of SI_SUB_KLD. Note that we don't believe that there is any consumers of x86bios that would be triggered in the early stage of boot, so this is more or less just a safebelt. Submitted by: swell.k at gmail.com
2009-09-21 09:09:55 +00:00
DECLARE_MODULE(x86bios, x86bios_mod, SI_SUB_CPU, SI_ORDER_ANY);
Collapase interrupt supporting functions to a new module, and switch from x86emu to this new module. This changeset also brings a fix for bugs introduced with the initial x86emu commit, which prevents the user from using some display mode or cause instant reboots during mode switch. Submitted by: paradox <ddkprog yahoo com>
2009-09-21 08:17:57 +00:00
MODULE_VERSION(x86bios, 1);
Reference in New Issue Copy Permalink