freebsd-nq/sys/boot/efi/loader/bootinfo.c

464 lines
13 KiB
C
Raw Normal View History

Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
/*-
* Copyright (c) 1998 Michael Smith <msmith@freebsd.org>
* Copyright (c) 2004, 2006 Marcel Moolenaar
* Copyright (c) 2014 The FreeBSD Foundation
* 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.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <stand.h>
#include <string.h>
#include <sys/param.h>
#include <sys/reboot.h>
#include <sys/linker.h>
#include <sys/boot.h>
#include <machine/cpufunc.h>
#include <machine/elf.h>
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
#include <machine/metadata.h>
#include <machine/psl.h>
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
#include <efi.h>
#include <efilib.h>
#include "bootstrap.h"
#include "loader_efi.h"
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
#if defined(__amd64__)
#include <machine/specialreg.h>
#include "framebuffer.h"
#endif
#if defined(LOADER_FDT_SUPPORT)
#include <fdt_platform.h>
#endif
int bi_load(char *args, vm_offset_t *modulep, vm_offset_t *kernendp);
extern EFI_SYSTEM_TABLE *ST;
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
static const char howto_switches[] = "aCdrgDmphsv";
static int howto_masks[] = {
RB_ASKNAME, RB_CDROM, RB_KDB, RB_DFLTROOT, RB_GDB, RB_MULTIPLE,
RB_MUTE, RB_PAUSE, RB_SERIAL, RB_SINGLE, RB_VERBOSE
};
static int
bi_getboothowto(char *kargs)
{
const char *sw;
char *opts;
char *console;
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
int howto, i;
howto = 0;
/* Get the boot options from the environment first. */
for (i = 0; howto_names[i].ev != NULL; i++) {
if (getenv(howto_names[i].ev) != NULL)
howto |= howto_names[i].mask;
}
console = getenv("console");
if (console != NULL) {
if (strcmp(console, "comconsole") == 0)
howto |= RB_SERIAL;
if (strcmp(console, "nullconsole") == 0)
howto |= RB_MUTE;
}
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
/* Parse kargs */
if (kargs == NULL)
return (howto);
opts = strchr(kargs, '-');
while (opts != NULL) {
while (*(++opts) != '\0') {
sw = strchr(howto_switches, *opts);
if (sw == NULL)
break;
howto |= howto_masks[sw - howto_switches];
}
opts = strchr(opts, '-');
}
return (howto);
}
/*
* Copy the environment into the load area starting at (addr).
* Each variable is formatted as <name>=<value>, with a single nul
* separating each variable, and a double nul terminating the environment.
*/
static vm_offset_t
bi_copyenv(vm_offset_t start)
{
struct env_var *ep;
vm_offset_t addr, last;
size_t len;
addr = last = start;
/* Traverse the environment. */
for (ep = environ; ep != NULL; ep = ep->ev_next) {
len = strlen(ep->ev_name);
if ((size_t)archsw.arch_copyin(ep->ev_name, addr, len) != len)
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
break;
addr += len;
if (archsw.arch_copyin("=", addr, 1) != 1)
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
break;
addr++;
if (ep->ev_value != NULL) {
len = strlen(ep->ev_value);
if ((size_t)archsw.arch_copyin(ep->ev_value, addr, len) != len)
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
break;
addr += len;
}
if (archsw.arch_copyin("", addr, 1) != 1)
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
break;
last = ++addr;
}
if (archsw.arch_copyin("", last++, 1) != 1)
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
last = start;
return(last);
}
/*
* Copy module-related data into the load area, where it can be
* used as a directory for loaded modules.
*
* Module data is presented in a self-describing format. Each datum
* is preceded by a 32-bit identifier and a 32-bit size field.
*
* Currently, the following data are saved:
*
* MOD_NAME (variable) module name (string)
* MOD_TYPE (variable) module type (string)
* MOD_ARGS (variable) module parameters (string)
* MOD_ADDR sizeof(vm_offset_t) module load address
* MOD_SIZE sizeof(size_t) module size
* MOD_METADATA (variable) type-specific metadata
*/
#define COPY32(v, a, c) { \
uint32_t x = (v); \
if (c) \
archsw.arch_copyin(&x, a, sizeof(x)); \
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
a += sizeof(x); \
}
#define MOD_STR(t, a, s, c) { \
COPY32(t, a, c); \
COPY32(strlen(s) + 1, a, c); \
if (c) \
archsw.arch_copyin(s, a, strlen(s) + 1); \
a += roundup(strlen(s) + 1, sizeof(u_long)); \
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
}
#define MOD_NAME(a, s, c) MOD_STR(MODINFO_NAME, a, s, c)
#define MOD_TYPE(a, s, c) MOD_STR(MODINFO_TYPE, a, s, c)
#define MOD_ARGS(a, s, c) MOD_STR(MODINFO_ARGS, a, s, c)
#define MOD_VAR(t, a, s, c) { \
COPY32(t, a, c); \
COPY32(sizeof(s), a, c); \
if (c) \
archsw.arch_copyin(&s, a, sizeof(s)); \
a += roundup(sizeof(s), sizeof(u_long)); \
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
}
#define MOD_ADDR(a, s, c) MOD_VAR(MODINFO_ADDR, a, s, c)
#define MOD_SIZE(a, s, c) MOD_VAR(MODINFO_SIZE, a, s, c)
#define MOD_METADATA(a, mm, c) { \
COPY32(MODINFO_METADATA | mm->md_type, a, c); \
COPY32(mm->md_size, a, c); \
if (c) \
archsw.arch_copyin(mm->md_data, a, mm->md_size); \
a += roundup(mm->md_size, sizeof(u_long)); \
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
}
#define MOD_END(a, c) { \
COPY32(MODINFO_END, a, c); \
COPY32(0, a, c); \
}
static vm_offset_t
bi_copymodules(vm_offset_t addr)
{
struct preloaded_file *fp;
struct file_metadata *md;
int c;
uint64_t v;
c = addr != 0;
/* Start with the first module on the list, should be the kernel. */
for (fp = file_findfile(NULL, NULL); fp != NULL; fp = fp->f_next) {
MOD_NAME(addr, fp->f_name, c); /* This must come first. */
MOD_TYPE(addr, fp->f_type, c);
if (fp->f_args)
MOD_ARGS(addr, fp->f_args, c);
v = fp->f_addr;
#if defined(__arm__)
v -= __elfN(relocation_offset);
#endif
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
MOD_ADDR(addr, v, c);
v = fp->f_size;
MOD_SIZE(addr, v, c);
for (md = fp->f_metadata; md != NULL; md = md->md_next)
if (!(md->md_type & MODINFOMD_NOCOPY))
MOD_METADATA(addr, md, c);
}
MOD_END(addr, c);
return(addr);
}
static int
bi_load_efi_data(struct preloaded_file *kfp)
{
EFI_MEMORY_DESCRIPTOR *mm;
EFI_PHYSICAL_ADDRESS addr;
EFI_STATUS status;
size_t efisz;
UINTN efi_mapkey;
UINTN mmsz, pages, retry, sz;
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
UINT32 mmver;
struct efi_map_header *efihdr;
#if defined(__amd64__)
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
struct efi_fb efifb;
if (efi_find_framebuffer(&efifb) == 0) {
printf("EFI framebuffer information:\n");
2014-09-18 13:00:03 +00:00
printf("addr, size 0x%lx, 0x%lx\n", efifb.fb_addr,
efifb.fb_size);
printf("dimensions %d x %d\n", efifb.fb_width,
efifb.fb_height);
printf("stride %d\n", efifb.fb_stride);
2014-09-18 13:00:03 +00:00
printf("masks 0x%08x, 0x%08x, 0x%08x, 0x%08x\n",
efifb.fb_mask_red, efifb.fb_mask_green, efifb.fb_mask_blue,
efifb.fb_mask_reserved);
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
file_addmetadata(kfp, MODINFOMD_EFI_FB, sizeof(efifb), &efifb);
}
#endif
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
efisz = (sizeof(struct efi_map_header) + 0xf) & ~0xf;
/*
* It is possible that the first call to ExitBootServices may change
* the map key. Fetch a new map key and retry ExitBootServices in that
* case.
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
*/
for (retry = 2; retry > 0; retry--) {
/*
* Allocate enough pages to hold the bootinfo block and the
* memory map EFI will return to us. The memory map has an
* unknown size, so we have to determine that first. Note that
* the AllocatePages call can itself modify the memory map, so
* we have to take that into account as well. The changes to
* the memory map are caused by splitting a range of free
* memory into two (AFAICT), so that one is marked as being
* loader data.
*/
sz = 0;
BS->GetMemoryMap(&sz, NULL, &efi_mapkey, &mmsz, &mmver);
sz += mmsz;
sz = (sz + 0xf) & ~0xf;
pages = EFI_SIZE_TO_PAGES(sz + efisz);
status = BS->AllocatePages(AllocateAnyPages, EfiLoaderData,
pages, &addr);
if (EFI_ERROR(status)) {
printf("%s: AllocatePages error %lu\n", __func__,
EFI_ERROR_CODE(status));
return (ENOMEM);
}
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
/*
* Read the memory map and stash it after bootinfo. Align the
* memory map on a 16-byte boundary (the bootinfo block is page
* aligned).
*/
efihdr = (struct efi_map_header *)addr;
mm = (void *)((uint8_t *)efihdr + efisz);
sz = (EFI_PAGE_SIZE * pages) - efisz;
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
status = BS->GetMemoryMap(&sz, mm, &efi_mapkey, &mmsz, &mmver);
if (EFI_ERROR(status)) {
printf("%s: GetMemoryMap error %lu\n", __func__,
EFI_ERROR_CODE(status));
return (EINVAL);
}
status = BS->ExitBootServices(IH, efi_mapkey);
if (EFI_ERROR(status) == 0) {
efihdr->memory_size = sz;
efihdr->descriptor_size = mmsz;
efihdr->descriptor_version = mmver;
file_addmetadata(kfp, MODINFOMD_EFI_MAP, efisz + sz,
efihdr);
return (0);
}
BS->FreePages(addr, pages);
}
printf("ExitBootServices error %lu\n", EFI_ERROR_CODE(status));
return (EINVAL);
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
}
/*
* Load the information expected by an amd64 kernel.
*
* - The 'boothowto' argument is constructed.
* - The 'bootdev' argument is constructed.
* - The 'bootinfo' struct is constructed, and copied into the kernel space.
* - The kernel environment is copied into kernel space.
* - Module metadata are formatted and placed in kernel space.
*/
int
bi_load(char *args, vm_offset_t *modulep, vm_offset_t *kernendp)
{
struct preloaded_file *xp, *kfp;
struct devdesc *rootdev;
struct file_metadata *md;
vm_offset_t addr;
uint64_t kernend;
uint64_t envp;
vm_offset_t size;
char *rootdevname;
int howto;
#if defined(LOADER_FDT_SUPPORT)
vm_offset_t dtbp;
int dtb_size;
#endif
#if defined(__arm__)
vm_offset_t vaddr;
size_t i;
/*
* These metadata addreses must be converted for kernel after
* relocation.
*/
uint32_t mdt[] = {
MODINFOMD_SSYM, MODINFOMD_ESYM, MODINFOMD_KERNEND,
MODINFOMD_ENVP,
#if defined(LOADER_FDT_SUPPORT)
MODINFOMD_DTBP
#endif
};
#endif
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
howto = bi_getboothowto(args);
/*
* Allow the environment variable 'rootdev' to override the supplied
* device. This should perhaps go to MI code and/or have $rootdev
* tested/set by MI code before launching the kernel.
*/
rootdevname = getenv("rootdev");
archsw.arch_getdev((void**)(&rootdev), rootdevname, NULL);
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
if (rootdev == NULL) {
printf("Can't determine root device.\n");
return(EINVAL);
}
/* Try reading the /etc/fstab file to select the root device */
getrootmount(efi_fmtdev((void *)rootdev));
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
addr = 0;
for (xp = file_findfile(NULL, NULL); xp != NULL; xp = xp->f_next) {
if (addr < (xp->f_addr + xp->f_size))
addr = xp->f_addr + xp->f_size;
}
/* Pad to a page boundary. */
addr = roundup(addr, PAGE_SIZE);
/* Copy our environment. */
envp = addr;
addr = bi_copyenv(addr);
/* Pad to a page boundary. */
addr = roundup(addr, PAGE_SIZE);
#if defined(LOADER_FDT_SUPPORT)
/* Handle device tree blob */
dtbp = addr;
dtb_size = fdt_copy(addr);
/* Pad to a page boundary */
if (dtb_size)
addr += roundup(dtb_size, PAGE_SIZE);
#endif
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
kfp = file_findfile(NULL, "elf kernel");
if (kfp == NULL)
kfp = file_findfile(NULL, "elf64 kernel");
if (kfp == NULL)
panic("can't find kernel file");
kernend = 0; /* fill it in later */
file_addmetadata(kfp, MODINFOMD_HOWTO, sizeof howto, &howto);
file_addmetadata(kfp, MODINFOMD_ENVP, sizeof envp, &envp);
#if defined(LOADER_FDT_SUPPORT)
if (dtb_size)
file_addmetadata(kfp, MODINFOMD_DTBP, sizeof dtbp, &dtbp);
else
pager_output("WARNING! Trying to fire up the kernel, but no "
"device tree blob found!\n");
#endif
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
file_addmetadata(kfp, MODINFOMD_KERNEND, sizeof kernend, &kernend);
file_addmetadata(kfp, MODINFOMD_FW_HANDLE, sizeof ST, &ST);
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
bi_load_efi_data(kfp);
/* Figure out the size and location of the metadata. */
*modulep = addr;
size = bi_copymodules(0);
kernend = roundup(addr + size, PAGE_SIZE);
*kernendp = kernend;
/* patch MODINFOMD_KERNEND */
md = file_findmetadata(kfp, MODINFOMD_KERNEND);
bcopy(&kernend, md->md_data, sizeof kernend);
#if defined(__arm__)
*modulep -= __elfN(relocation_offset);
/* Do relocation fixup on metadata of each module. */
for (xp = file_findfile(NULL, NULL); xp != NULL; xp = xp->f_next) {
for (i = 0; i < sizeof mdt / sizeof mdt[0]; i++) {
md = file_findmetadata(xp, mdt[i]);
if (md) {
bcopy(md->md_data, &vaddr, sizeof vaddr);
vaddr -= __elfN(relocation_offset);
bcopy(&vaddr, md->md_data, sizeof vaddr);
}
}
}
#endif
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
/* Copy module list and metadata. */
(void)bi_copymodules(addr);
Support UEFI booting on amd64 via loader.efi This is largely the work from the projects/uefi branch, with some additional refinements. This is derived from (and replaces) the original i386 efi implementation; i386 support will be restored later. Specific revisions of note from projects/uefi: r247380: Adjust our load device when we boot from CD under UEFI. The process for booting from a CD under UEFI involves adding a FAT filesystem containing your loader code as an El Torito boot image. When UEFI detects this, it provides a block IO instance that points at the FAT filesystem as a child of the device that represents the CD itself. The problem being that the CD device is flagged as a "raw device" while the boot image is flagged as a "logical partition". The existing EFI partition code only looks for logical partitions and so the CD filesystem was rendered invisible. To fix this, check the type of each block IO device. If it's found to be a CD, and thus an El Torito boot image, look up its parent device and add that instead so that the loader will then load the kernel from the CD filesystem. This is done by using the handle for the boot filesystem as an alias. Something similar to this will be required for booting from other media as well as the loader will live in the EFI system partition, not on the partition containing the kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r246231: Add necessary code to hand off from loader to an amd64 kernel. r246335: Grab the EFI memory map and store it as module metadata on the kernel. This is the same approach used to provide the BIOS SMAP to the kernel. r246336: Pass the ACPI table metadata via hints so the kernel ACPI code can find them. r246608: Rework copy routines to ensure we always use memory allocated via EFI. The previous code assumed it could copy wherever it liked. This is not the case. The approach taken by this code is pretty ham-fisted in that it simply allocates a large (32MB) buffer area and stages into that, then copies the whole area into place when it's time to execute. A more elegant solution could be used but this works for now. r247214: Fix a number of problems preventing proper handover to the kernel. There were two issues at play here. Firstly, there was nothing preventing UEFI from placing the loader code above 1GB in RAM. This meant that when we switched in the page tables the kernel expects to be running on, we are suddenly unmapped and things no longer work. We solve this by making our trampoline code not dependent on being at any given position and simply copying it to a "safe" location before calling it. Secondly, UEFI could allocate our stack wherever it wants. As it happened on my PC, that was right where I was copying the kernel to. This did not cause happiness. The solution to this was to also switch to a temporary stack in a safe location before performing the final copy of the loaded kernel. r247216: Use the UEFI Graphics Output Protocol to get the parameters of the framebuffer. Sponsored by: The FreeBSD Foundation
2014-04-04 00:16:46 +00:00
return (0);
}