freebsd-nq/sys/dev/ti/if_tireg.h

1101 lines
33 KiB
C
Raw Normal View History

/*-
* Copyright (c) 1997, 1998, 1999
* Bill Paul <wpaul@ctr.columbia.edu>. 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.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by Bill Paul.
* 4. Neither the name of the author nor the names of any co-contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY Bill Paul 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 Bill Paul OR THE VOICES IN HIS HEAD
* 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.
*
1999-08-28 01:08:13 +00:00
* $FreeBSD$
*/
/*
* Tigon register offsets. These are memory mapped registers
* which can be accessed with the CSR_READ_4()/CSR_WRITE_4() macros.
* Each register must be accessed using 32 bit operations.
*
* All reegisters are accessed through a 16K shared memory block.
* The first group of registers are actually copies of the PCI
* configuration space registers.
*/
#define TI_PCI_ID 0x000 /* PCI device/vendor ID */
#define TI_PCI_CMDSTAT 0x004
#define TI_PCI_CLASSCODE 0x008
#define TI_PCI_BIST 0x00C
#define TI_PCI_LOMEM 0x010 /* Shared memory base address */
#define TI_PCI_SUBSYS 0x02C
#define TI_PCI_ROMBASE 0x030
#define TI_PCI_INT 0x03C
#ifndef PCIM_CMD_MWIEN
#define PCIM_CMD_MWIEN 0x0010
#endif
/*
* Alteon AceNIC PCI vendor/device ID.
*/
#define ALT_VENDORID 0x12AE
#define ALT_DEVICEID_ACENIC 0x0001
#define ALT_DEVICEID_ACENIC_COPPER 0x0002
/*
* 3Com 3c985 PCI vendor/device ID.
*/
#define TC_VENDORID 0x10B7
#define TC_DEVICEID_3C985 0x0001
/*
* Netgear GA620 PCI vendor/device ID.
*/
#define NG_VENDORID 0x1385
#define NG_DEVICEID_GA620 0x620A
#define NG_DEVICEID_GA620T 0x630A
/*
* SGI device/vendor ID.
*/
#define SGI_VENDORID 0x10A9
#define SGI_DEVICEID_TIGON 0x0009
/*
* DEC vendor ID, Farallon device ID. Apparently, Farallon used
* the DEC vendor ID in their cards by mistake.
*/
#define DEC_VENDORID 0x1011
#define DEC_DEVICEID_FARALLON_PN9000SX 0x001a
/*
* Tigon configuration and control registers.
*/
#define TI_MISC_HOST_CTL 0x040
#define TI_MISC_LOCAL_CTL 0x044
#define TI_SEM_AB 0x048 /* Tigon 2 only */
#define TI_MISC_CONF 0x050 /* Tigon 2 only */
#define TI_TIMER_BITS 0x054
#define TI_TIMERREF 0x058
#define TI_PCI_STATE 0x05C
#define TI_MAIN_EVENT_A 0x060
#define TI_MAILBOX_EVENT_A 0x064
#define TI_WINBASE 0x068
#define TI_WINDATA 0x06C
#define TI_MAIN_EVENT_B 0x070 /* Tigon 2 only */
#define TI_MAILBOX_EVENT_B 0x074 /* Tigon 2 only */
#define TI_TIMERREF_B 0x078 /* Tigon 2 only */
#define TI_SERIAL 0x07C
/*
* Misc host control bits.
*/
#define TI_MHC_INTSTATE 0x00000001
#define TI_MHC_CLEARINT 0x00000002
#define TI_MHC_RESET 0x00000008
#define TI_MHC_BYTE_SWAP_ENB 0x00000010
#define TI_MHC_WORD_SWAP_ENB 0x00000020
#define TI_MHC_MASK_INTS 0x00000040
#define TI_MHC_CHIP_REV_MASK 0xF0000000
#define TI_MHC_BIGENDIAN_INIT \
(TI_MHC_BYTE_SWAP_ENB|TI_MHC_WORD_SWAP_ENB|TI_MHC_CLEARINT)
#define TI_MHC_LITTLEENDIAN_INIT \
(TI_MHC_WORD_SWAP_ENB|TI_MHC_CLEARINT)
/*
* Tigon chip rev values. Rev 4 is the Tigon 1. Rev 6 is the Tigon 2.
* Rev 5 is also the Tigon 2, but is a broken version which was never
* used in any actual hardware, so we ignore it.
*/
#define TI_REV_TIGON_I 0x40000000
#define TI_REV_TIGON_II 0x60000000
/*
* Firmware revision that we want.
*/
#define TI_FIRMWARE_MAJOR 0xc
#define TI_FIRMWARE_MINOR 0x4
At long last, commit the zero copy sockets code. MAKEDEV: Add MAKEDEV glue for the ti(4) device nodes. ti.4: Update the ti(4) man page to include information on the TI_JUMBO_HDRSPLIT and TI_PRIVATE_JUMBOS kernel options, and also include information about the new character device interface and the associated ioctls. man9/Makefile: Add jumbo.9 and zero_copy.9 man pages and associated links. jumbo.9: New man page describing the jumbo buffer allocator interface and operation. zero_copy.9: New man page describing the general characteristics of the zero copy send and receive code, and what an application author should do to take advantage of the zero copy functionality. NOTES: Add entries for ZERO_COPY_SOCKETS, TI_PRIVATE_JUMBOS, TI_JUMBO_HDRSPLIT, MSIZE, and MCLSHIFT. conf/files: Add uipc_jumbo.c and uipc_cow.c. conf/options: Add the 5 options mentioned above. kern_subr.c: Receive side zero copy implementation. This takes "disposable" pages attached to an mbuf, gives them to a user process, and then recycles the user's page. This is only active when ZERO_COPY_SOCKETS is turned on and the kern.ipc.zero_copy.receive sysctl variable is set to 1. uipc_cow.c: Send side zero copy functions. Takes a page written by the user and maps it copy on write and assigns it kernel virtual address space. Removes copy on write mapping once the buffer has been freed by the network stack. uipc_jumbo.c: Jumbo disposable page allocator code. This allocates (optionally) disposable pages for network drivers that want to give the user the option of doing zero copy receive. uipc_socket.c: Add kern.ipc.zero_copy.{send,receive} sysctls that are enabled if ZERO_COPY_SOCKETS is turned on. Add zero copy send support to sosend() -- pages get mapped into the kernel instead of getting copied if they meet size and alignment restrictions. uipc_syscalls.c:Un-staticize some of the sf* functions so that they can be used elsewhere. (uipc_cow.c) if_media.c: In the SIOCGIFMEDIA ioctl in ifmedia_ioctl(), avoid calling malloc() with M_WAITOK. Return an error if the M_NOWAIT malloc fails. The ti(4) driver and the wi(4) driver, at least, call this with a mutex held. This causes witness warnings for 'ifconfig -a' with a wi(4) or ti(4) board in the system. (I've only verified for ti(4)). ip_output.c: Fragment large datagrams so that each segment contains a multiple of PAGE_SIZE amount of data plus headers. This allows the receiver to potentially do page flipping on receives. if_ti.c: Add zero copy receive support to the ti(4) driver. If TI_PRIVATE_JUMBOS is not defined, it now uses the jumbo(9) buffer allocator for jumbo receive buffers. Add a new character device interface for the ti(4) driver for the new debugging interface. This allows (a patched version of) gdb to talk to the Tigon board and debug the firmware. There are also a few additional debugging ioctls available through this interface. Add header splitting support to the ti(4) driver. Tweak some of the default interrupt coalescing parameters to more useful defaults. Add hooks for supporting transmit flow control, but leave it turned off with a comment describing why it is turned off. if_tireg.h: Change the firmware rev to 12.4.11, since we're really at 12.4.11 plus fixes from 12.4.13. Add defines needed for debugging. Remove the ti_stats structure, it is now defined in sys/tiio.h. ti_fw.h: 12.4.11 firmware. ti_fw2.h: 12.4.11 firmware, plus selected fixes from 12.4.13, and my header splitting patches. Revision 12.4.13 doesn't handle 10/100 negotiation properly. (This firmware is the same as what was in the tree previously, with the addition of header splitting support.) sys/jumbo.h: Jumbo buffer allocator interface. sys/mbuf.h: Add a new external mbuf type, EXT_DISPOSABLE, to indicate that the payload buffer can be thrown away / flipped to a userland process. socketvar.h: Add prototype for socow_setup. tiio.h: ioctl interface to the character portion of the ti(4) driver, plus associated structure/type definitions. uio.h: Change prototype for uiomoveco() so that we'll know whether the source page is disposable. ufs_readwrite.c:Update for new prototype of uiomoveco(). vm_fault.c: In vm_fault(), check to see whether we need to do a page based copy on write fault. vm_object.c: Add a new function, vm_object_allocate_wait(). This does the same thing that vm_object allocate does, except that it gives the caller the opportunity to specify whether it should wait on the uma_zalloc() of the object structre. This allows vm objects to be allocated while holding a mutex. (Without generating WITNESS warnings.) vm_object_allocate() is implemented as a call to vm_object_allocate_wait() with the malloc flag set to M_WAITOK. vm_object.h: Add prototype for vm_object_allocate_wait(). vm_page.c: Add page-based copy on write setup, clear and fault routines. vm_page.h: Add page based COW function prototypes and variable in the vm_page structure. Many thanks to Drew Gallatin, who wrote the zero copy send and receive code, and to all the other folks who have tested and reviewed this code over the years.
2002-06-26 03:37:47 +00:00
#define TI_FIRMWARE_FIX 0xb
/*
* Miscelaneous Local Control register.
*/
#define TI_MLC_EE_WRITE_ENB 0x00000010
#define TI_MLC_SRAM_BANK_SIZE 0x00000300 /* Tigon 2 only */
#define TI_MLC_LOCALADDR_21 0x00004000
#define TI_MLC_LOCALADDR_22 0x00008000
#define TI_MLC_SBUS_WRITEERR 0x00080000
#define TI_MLC_EE_CLK 0x00100000
#define TI_MLC_EE_TXEN 0x00200000
#define TI_MLC_EE_DOUT 0x00400000
#define TI_MLC_EE_DIN 0x00800000
/* Possible memory sizes. */
#define TI_MLC_SRAM_BANK_DISA 0x00000000
#define TI_MLC_SRAM_BANK_1024K 0x00000100
#define TI_MLC_SRAM_BANK_512K 0x00000200
#define TI_MLC_SRAM_BANK_256K 0x00000300
/*
* Offset of MAC address inside EEPROM.
*/
#define TI_EE_MAC_OFFSET 0x8c
#define TI_DMA_ASSIST 0x11C
#define TI_CPU_STATE 0x140
#define TI_CPU_PROGRAM_COUNTER 0x144
#define TI_SRAM_ADDR 0x154
#define TI_SRAM_DATA 0x158
#define TI_GEN_0 0x180
#define TI_GEN_X 0x1FC
#define TI_MAC_TX_STATE 0x200
#define TI_MAC_RX_STATE 0x220
#define TI_CPU_CTL_B 0x240 /* Tigon 2 only */
#define TI_CPU_PROGRAM_COUNTER_B 0x244 /* Tigon 2 only */
#define TI_SRAM_ADDR_B 0x254 /* Tigon 2 only */
#define TI_SRAM_DATA_B 0x258 /* Tigon 2 only */
#define TI_GEN_B_0 0x280 /* Tigon 2 only */
#define TI_GEN_B_X 0x2FC /* Tigon 2 only */
/*
* Misc config register.
*/
#define TI_MCR_SRAM_SYNCHRONOUS 0x00100000 /* Tigon 2 only */
/*
* PCI state register.
*/
#define TI_PCISTATE_FORCE_RESET 0x00000001
#define TI_PCISTATE_PROVIDE_LEN 0x00000002
#define TI_PCISTATE_READ_MAXDMA 0x0000001C
#define TI_PCISTATE_WRITE_MAXDMA 0x000000E0
#define TI_PCISTATE_MINDMA 0x0000FF00
#define TI_PCISTATE_FIFO_RETRY_ENB 0x00010000
#define TI_PCISTATE_USE_MEM_RD_MULT 0x00020000
#define TI_PCISTATE_NO_SWAP_READ_DMA 0x00040000
#define TI_PCISTATE_NO_SWAP_WRITE_DMA 0x00080000
#define TI_PCISTATE_66MHZ_BUS 0x00080000 /* Tigon 2 only */
#define TI_PCISTATE_32BIT_BUS 0x00100000 /* Tigon 2 only */
#define TI_PCISTATE_ENB_BYTE_ENABLES 0x00800000 /* Tigon 2 only */
#define TI_PCISTATE_READ_CMD 0x0F000000
#define TI_PCISTATE_WRITE_CMD 0xF0000000
#define TI_PCI_READMAX_4 0x04
#define TI_PCI_READMAX_16 0x08
#define TI_PCI_READMAX_32 0x0C
#define TI_PCI_READMAX_64 0x10
#define TI_PCI_READMAX_128 0x14
#define TI_PCI_READMAX_256 0x18
#define TI_PCI_READMAX_1024 0x1C
#define TI_PCI_WRITEMAX_4 0x20
#define TI_PCI_WRITEMAX_16 0x40
#define TI_PCI_WRITEMAX_32 0x60
#define TI_PCI_WRITEMAX_64 0x80
#define TI_PCI_WRITEMAX_128 0xA0
#define TI_PCI_WRITEMAX_256 0xC0
#define TI_PCI_WRITEMAX_1024 0xE0
#define TI_PCI_READ_CMD 0x06000000
#define TI_PCI_WRITE_CMD 0x70000000
/*
* DMA state register.
*/
#define TI_DMASTATE_ENABLE 0x00000001
#define TI_DMASTATE_PAUSE 0x00000002
/*
* CPU state register.
*/
#define TI_CPUSTATE_RESET 0x00000001
#define TI_CPUSTATE_STEP 0x00000002
#define TI_CPUSTATE_ROMFAIL 0x00000010
#define TI_CPUSTATE_HALT 0x00010000
/*
* MAC TX state register
*/
#define TI_TXSTATE_RESET 0x00000001
#define TI_TXSTATE_ENB 0x00000002
#define TI_TXSTATE_STOP 0x00000004
/*
* MAC RX state register
*/
#define TI_RXSTATE_RESET 0x00000001
#define TI_RXSTATE_ENB 0x00000002
#define TI_RXSTATE_STOP 0x00000004
/*
* Tigon 2 mailbox registers. The mailbox area consists of 256 bytes
* split into 64 bit registers. Only the lower 32 bits of each mailbox
* are used.
*/
#define TI_MB_HOSTINTR_HI 0x500
#define TI_MB_HOSTINTR_LO 0x504
#define TI_MB_HOSTINTR TI_MB_HOSTINTR_LO
#define TI_MB_CMDPROD_IDX_HI 0x508
#define TI_MB_CMDPROD_IDX_LO 0x50C
#define TI_MB_CMDPROD_IDX TI_MB_CMDPROD_IDX_LO
#define TI_MB_SENDPROD_IDX_HI 0x510
#define TI_MB_SENDPROD_IDX_LO 0x514
#define TI_MB_SENDPROD_IDX TI_MB_SENDPROD_IDX_LO
#define TI_MB_STDRXPROD_IDX_HI 0x518 /* Tigon 2 only */
#define TI_MB_STDRXPROD_IDX_LO 0x51C /* Tigon 2 only */
#define TI_MB_STDRXPROD_IDX TI_MB_STDRXPROD_IDX_LO
#define TI_MB_JUMBORXPROD_IDX_HI 0x520 /* Tigon 2 only */
#define TI_MB_JUMBORXPROD_IDX_LO 0x524 /* Tigon 2 only */
#define TI_MB_JUMBORXPROD_IDX TI_MB_JUMBORXPROD_IDX_LO
#define TI_MB_MINIRXPROD_IDX_HI 0x528 /* Tigon 2 only */
#define TI_MB_MINIRXPROD_IDX_LO 0x52C /* Tigon 2 only */
#define TI_MB_MINIRXPROD_IDX TI_MB_MINIRXPROD_IDX_LO
#define TI_MB_RSVD 0x530
/*
* Tigon 2 general communication registers. These are 64 and 32 bit
* registers which are only valid after the firmware has been
* loaded and started. They actually exist in NIC memory but are
* mapped into the host memory via the shared memory region.
*
* The NIC internally maps these registers starting at address 0,
* so to determine the NIC address of any of these registers, we
* subtract 0x600 (the address of the first register).
*/
#define TI_GCR_BASE 0x600
#define TI_GCR_MACADDR 0x600
#define TI_GCR_PAR0 0x600
#define TI_GCR_PAR1 0x604
#define TI_GCR_GENINFO_HI 0x608
#define TI_GCR_GENINFO_LO 0x60C
#define TI_GCR_MCASTADDR 0x610 /* obsolete */
#define TI_GCR_MAR0 0x610 /* obsolete */
#define TI_GCR_MAR1 0x614 /* obsolete */
#define TI_GCR_OPMODE 0x618
#define TI_GCR_DMA_READCFG 0x61C
#define TI_GCR_DMA_WRITECFG 0x620
#define TI_GCR_TX_BUFFER_RATIO 0x624
#define TI_GCR_EVENTCONS_IDX 0x628
#define TI_GCR_CMDCONS_IDX 0x62C
#define TI_GCR_TUNEPARMS 0x630
#define TI_GCR_RX_COAL_TICKS 0x630
#define TI_GCR_TX_COAL_TICKS 0x634
#define TI_GCR_STAT_TICKS 0x638
#define TI_GCR_TX_MAX_COAL_BD 0x63C
#define TI_GCR_RX_MAX_COAL_BD 0x640
#define TI_GCR_NIC_TRACING 0x644
#define TI_GCR_GLINK 0x648
#define TI_GCR_LINK 0x64C
#define TI_GCR_NICTRACE_PTR 0x650
#define TI_GCR_NICTRACE_START 0x654
#define TI_GCR_NICTRACE_LEN 0x658
#define TI_GCR_IFINDEX 0x65C
#define TI_GCR_IFMTU 0x660
#define TI_GCR_MASK_INTRS 0x664
#define TI_GCR_GLINK_STAT 0x668
#define TI_GCR_LINK_STAT 0x66C
#define TI_GCR_RXRETURNCONS_IDX 0x680
#define TI_GCR_CMDRING 0x700
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
#define TI_GCR_NIC_ADDR(x) (x - TI_GCR_BASE)
/*
* Local memory window. The local memory window is a 2K shared
* memory region which can be used to access the NIC's internal
* SRAM. The window can be mapped to a given 2K region using
* the TI_WINDOW_BASE register.
*/
#define TI_WINDOW 0x800
#define TI_WINLEN 0x800
#define TI_TICKS_PER_SEC 1000000
/*
* Operation mode register.
*/
#define TI_OPMODE_BYTESWAP_BD 0x00000002
#define TI_OPMODE_WORDSWAP_BD 0x00000004
#define TI_OPMODE_WARN_ENB 0x00000008 /* not yet implimented */
#define TI_OPMODE_BYTESWAP_DATA 0x00000010
#define TI_OPMODE_1_DMA_ACTIVE 0x00000040
#define TI_OPMODE_SBUS 0x00000100
#define TI_OPMODE_DONT_FRAG_JUMBO 0x00000200
#define TI_OPMODE_INCLUDE_CRC 0x00000400
#define TI_OPMODE_RX_BADFRAMES 0x00000800
#define TI_OPMODE_NO_EVENT_INTRS 0x00001000
#define TI_OPMODE_NO_TX_INTRS 0x00002000
#define TI_OPMODE_NO_RX_INTRS 0x00004000
#define TI_OPMODE_FATAL_ENB 0x40000000 /* not yet implimented */
At long last, commit the zero copy sockets code. MAKEDEV: Add MAKEDEV glue for the ti(4) device nodes. ti.4: Update the ti(4) man page to include information on the TI_JUMBO_HDRSPLIT and TI_PRIVATE_JUMBOS kernel options, and also include information about the new character device interface and the associated ioctls. man9/Makefile: Add jumbo.9 and zero_copy.9 man pages and associated links. jumbo.9: New man page describing the jumbo buffer allocator interface and operation. zero_copy.9: New man page describing the general characteristics of the zero copy send and receive code, and what an application author should do to take advantage of the zero copy functionality. NOTES: Add entries for ZERO_COPY_SOCKETS, TI_PRIVATE_JUMBOS, TI_JUMBO_HDRSPLIT, MSIZE, and MCLSHIFT. conf/files: Add uipc_jumbo.c and uipc_cow.c. conf/options: Add the 5 options mentioned above. kern_subr.c: Receive side zero copy implementation. This takes "disposable" pages attached to an mbuf, gives them to a user process, and then recycles the user's page. This is only active when ZERO_COPY_SOCKETS is turned on and the kern.ipc.zero_copy.receive sysctl variable is set to 1. uipc_cow.c: Send side zero copy functions. Takes a page written by the user and maps it copy on write and assigns it kernel virtual address space. Removes copy on write mapping once the buffer has been freed by the network stack. uipc_jumbo.c: Jumbo disposable page allocator code. This allocates (optionally) disposable pages for network drivers that want to give the user the option of doing zero copy receive. uipc_socket.c: Add kern.ipc.zero_copy.{send,receive} sysctls that are enabled if ZERO_COPY_SOCKETS is turned on. Add zero copy send support to sosend() -- pages get mapped into the kernel instead of getting copied if they meet size and alignment restrictions. uipc_syscalls.c:Un-staticize some of the sf* functions so that they can be used elsewhere. (uipc_cow.c) if_media.c: In the SIOCGIFMEDIA ioctl in ifmedia_ioctl(), avoid calling malloc() with M_WAITOK. Return an error if the M_NOWAIT malloc fails. The ti(4) driver and the wi(4) driver, at least, call this with a mutex held. This causes witness warnings for 'ifconfig -a' with a wi(4) or ti(4) board in the system. (I've only verified for ti(4)). ip_output.c: Fragment large datagrams so that each segment contains a multiple of PAGE_SIZE amount of data plus headers. This allows the receiver to potentially do page flipping on receives. if_ti.c: Add zero copy receive support to the ti(4) driver. If TI_PRIVATE_JUMBOS is not defined, it now uses the jumbo(9) buffer allocator for jumbo receive buffers. Add a new character device interface for the ti(4) driver for the new debugging interface. This allows (a patched version of) gdb to talk to the Tigon board and debug the firmware. There are also a few additional debugging ioctls available through this interface. Add header splitting support to the ti(4) driver. Tweak some of the default interrupt coalescing parameters to more useful defaults. Add hooks for supporting transmit flow control, but leave it turned off with a comment describing why it is turned off. if_tireg.h: Change the firmware rev to 12.4.11, since we're really at 12.4.11 plus fixes from 12.4.13. Add defines needed for debugging. Remove the ti_stats structure, it is now defined in sys/tiio.h. ti_fw.h: 12.4.11 firmware. ti_fw2.h: 12.4.11 firmware, plus selected fixes from 12.4.13, and my header splitting patches. Revision 12.4.13 doesn't handle 10/100 negotiation properly. (This firmware is the same as what was in the tree previously, with the addition of header splitting support.) sys/jumbo.h: Jumbo buffer allocator interface. sys/mbuf.h: Add a new external mbuf type, EXT_DISPOSABLE, to indicate that the payload buffer can be thrown away / flipped to a userland process. socketvar.h: Add prototype for socow_setup. tiio.h: ioctl interface to the character portion of the ti(4) driver, plus associated structure/type definitions. uio.h: Change prototype for uiomoveco() so that we'll know whether the source page is disposable. ufs_readwrite.c:Update for new prototype of uiomoveco(). vm_fault.c: In vm_fault(), check to see whether we need to do a page based copy on write fault. vm_object.c: Add a new function, vm_object_allocate_wait(). This does the same thing that vm_object allocate does, except that it gives the caller the opportunity to specify whether it should wait on the uma_zalloc() of the object structre. This allows vm objects to be allocated while holding a mutex. (Without generating WITNESS warnings.) vm_object_allocate() is implemented as a call to vm_object_allocate_wait() with the malloc flag set to M_WAITOK. vm_object.h: Add prototype for vm_object_allocate_wait(). vm_page.c: Add page-based copy on write setup, clear and fault routines. vm_page.h: Add page based COW function prototypes and variable in the vm_page structure. Many thanks to Drew Gallatin, who wrote the zero copy send and receive code, and to all the other folks who have tested and reviewed this code over the years.
2002-06-26 03:37:47 +00:00
#define TI_OPMODE_JUMBO_HDRSPLIT 0x00008000
/*
* DMA configuration thresholds.
*/
#define TI_DMA_STATE_THRESH_16W 0x00000100
#define TI_DMA_STATE_THRESH_8W 0x00000080
#define TI_DMA_STATE_THRESH_4W 0x00000040
#define TI_DMA_STATE_THRESH_2W 0x00000020
#define TI_DMA_STATE_THRESH_1W 0x00000010
#define TI_DMA_STATE_FORCE_32_BIT 0x00000008
/*
* Gigabit link status bits.
*/
#define TI_GLNK_SENSE_NO_BEG 0x00002000
#define TI_GLNK_LOOPBACK 0x00004000
#define TI_GLNK_PREF 0x00008000
#define TI_GLNK_1000MB 0x00040000
#define TI_GLNK_FULL_DUPLEX 0x00080000
#define TI_GLNK_TX_FLOWCTL_Y 0x00200000 /* Tigon 2 only */
#define TI_GLNK_RX_FLOWCTL_Y 0x00800000
#define TI_GLNK_AUTONEGENB 0x20000000
#define TI_GLNK_ENB 0x40000000
/*
* Link status bits.
*/
#define TI_LNK_LOOPBACK 0x00004000
#define TI_LNK_PREF 0x00008000
#define TI_LNK_10MB 0x00010000
#define TI_LNK_100MB 0x00020000
#define TI_LNK_1000MB 0x00040000
#define TI_LNK_FULL_DUPLEX 0x00080000
#define TI_LNK_HALF_DUPLEX 0x00100000
#define TI_LNK_TX_FLOWCTL_Y 0x00200000 /* Tigon 2 only */
#define TI_LNK_RX_FLOWCTL_Y 0x00800000
#define TI_LNK_AUTONEGENB 0x20000000
#define TI_LNK_ENB 0x40000000
/*
* Ring size constants.
*/
#define TI_EVENT_RING_CNT 256
#define TI_CMD_RING_CNT 64
#define TI_STD_RX_RING_CNT 512
#define TI_JUMBO_RX_RING_CNT 256
#define TI_MINI_RX_RING_CNT 1024
#define TI_RETURN_RING_CNT 2048
- Tx side bus_dmamap_load_mbuf_sg(9) support. This reduces bookkeeping requiried to keep consistent softc state before/after callback function invocation and supposed to be sligntly faster than previous one as it wouldn't incur callback overhead. With this change callback function was gone. - Decrease TI_MAXTXSEGS to 32 from 128. It seems that most mbuf chain length is less than 32 and it would be re-packed with m_defrag(9) if its chain length is larger than TI_MAXTXSEGS. This would protect ti(4) against possible kernel stack overflow when txsegs[] is put on stack. Alternatively, we can embed the txsegs[] into softc. However, that would waste memory and make Tx/Rx speration hard when we want to sperate Tx/Rx handlers to optimize locking. - Fix dma map tracking used in Tx path. Previously it used the dma map of the last mbuf chain in ti_txeof() which was incorrect as ti(4) used dma map of the first mbuf chain when it loads a mbuf chain with bus_dmamap_load_mbuf(9). Correct the bug by introducing queues that keep track of active/inactive dma maps/mbuf chain. - Use ti_txcnt to check whether driver need to set watchdog timer instead of blidnly clearing the timer in ti_txeof(). - Remove the 3rd arg. of ti_encap(). Since ti(4) now caches the last descriptor index(ti_tx_saved_prodidx) used in Tx there is no need to pass it as a fuction arg. - Change data type of producer/consumer index to int from u_int16_t in order to remove implicit type conversions in Tx/Rx handlers. - Check interface queue before getting a mbuf chain to reduce locking overhead. - Check number of available Tx descriptores to be 16 or higher in ti_start(). This wouldn't protect Tx descriptor shortage but it would reduce number of bus_dmamap_unload(9) calls in ti_encap() when we are about to running out of Tx descriptors. - Command NIC to send packets ony when the driver really has packets enqueued. Previously it always set TI_MB_SENDPROD_IDX which would command NIC to DMA Tx descriptors into NIC local memory regardless of Tx descriptor changes. Reviewed by: scottl
2006-01-03 06:14:07 +00:00
#define TI_MAXTXSEGS 32
Overhaul bus_dma(9) usage in driver: - Don't use a single big DMA block for all rings. Create separate DMA area for each ring instead. Currently the following DMA areas are created: Event ring, standard RX ring, jumbo RX ring, RX return ring, hardware MAC statistics and producer/consumer status area. For Tigon II, mini RX ring and TX ring are additionally created. - Added missing bus_dmamap_sync(9) in various TX/RX paths. - TX ring is no longer created for Tigon 1 such that it saves more resources on Tigon 1. - Data sheet is not clear about alignment requirement of each ring so use 32 bytes alignment for normal DMA area but use 64 bytes alignment for jumbo RX ring where the extended RX descriptor size is 64 bytes. - For each TX/RX buffers use separate DMA tag(e.g. the size of a DMA segment, total size of DMA segments etc). - Tigon allows separate DMA area for event producer, RX return producer and TX consumer which is really cool feature. This means TX and RX path could be independently run in parallel. However ti(4) uses a single driver lock so it's meaningless to have separate DMA area for these producer/consumer such that this change creates a single status DMA area. - It seems Tigon has no limits on DMA address space and I also don't see any problem with that but old comments in driver indicates there could be issues on descriptors being located in 64bit region. Introduce a tunable, dev.ti.%d.dac, to disable using 64bit DMA in driver. The default is 0 which means it would use full 64bit DMA. If there are DMA issues, users can disable it by setting the tunable to 0. - Do not increase watchdog timer in ti_txeof(). Previously driver increased the watchdog timer whenever there are queued TX frames. - When stat ticks is set to 0, skip processing ti_stats_update(), avoiding bus_dmamap_sync(9) and updating if_collisions counter. - MTU does not include FCS bytes, replace it with ETHER_VLAN_ENCAP_LEN. With these changes, ti(4) should work on PAE environments. Many thanks to Jay Borkenhagen for remote hardware access.
2011-11-14 20:38:14 +00:00
#define TI_RING_ALIGN 32
#define TI_JUMBO_RING_ALIGN 64
/*
* Possible TX ring sizes.
*/
#define TI_TX_RING_CNT_128 128
#define TI_TX_RING_BASE_128 0x3800
#define TI_TX_RING_CNT_256 256
#define TI_TX_RING_BASE_256 0x3000
#define TI_TX_RING_CNT_512 512
#define TI_TX_RING_BASE_512 0x2000
#define TI_TX_RING_CNT TI_TX_RING_CNT_512
#define TI_TX_RING_BASE TI_TX_RING_BASE_512
/*
* The Tigon can have up to 8MB of external SRAM, however the Tigon 1
* is limited to 2MB total, and in general I think most adapters have
* around 1MB. We use this value for zeroing the NIC's SRAM, so to
* be safe we use the largest possible value (zeroing memory that
* isn't there doesn't hurt anything).
*/
#define TI_MEM_MAX 0x7FFFFF
At long last, commit the zero copy sockets code. MAKEDEV: Add MAKEDEV glue for the ti(4) device nodes. ti.4: Update the ti(4) man page to include information on the TI_JUMBO_HDRSPLIT and TI_PRIVATE_JUMBOS kernel options, and also include information about the new character device interface and the associated ioctls. man9/Makefile: Add jumbo.9 and zero_copy.9 man pages and associated links. jumbo.9: New man page describing the jumbo buffer allocator interface and operation. zero_copy.9: New man page describing the general characteristics of the zero copy send and receive code, and what an application author should do to take advantage of the zero copy functionality. NOTES: Add entries for ZERO_COPY_SOCKETS, TI_PRIVATE_JUMBOS, TI_JUMBO_HDRSPLIT, MSIZE, and MCLSHIFT. conf/files: Add uipc_jumbo.c and uipc_cow.c. conf/options: Add the 5 options mentioned above. kern_subr.c: Receive side zero copy implementation. This takes "disposable" pages attached to an mbuf, gives them to a user process, and then recycles the user's page. This is only active when ZERO_COPY_SOCKETS is turned on and the kern.ipc.zero_copy.receive sysctl variable is set to 1. uipc_cow.c: Send side zero copy functions. Takes a page written by the user and maps it copy on write and assigns it kernel virtual address space. Removes copy on write mapping once the buffer has been freed by the network stack. uipc_jumbo.c: Jumbo disposable page allocator code. This allocates (optionally) disposable pages for network drivers that want to give the user the option of doing zero copy receive. uipc_socket.c: Add kern.ipc.zero_copy.{send,receive} sysctls that are enabled if ZERO_COPY_SOCKETS is turned on. Add zero copy send support to sosend() -- pages get mapped into the kernel instead of getting copied if they meet size and alignment restrictions. uipc_syscalls.c:Un-staticize some of the sf* functions so that they can be used elsewhere. (uipc_cow.c) if_media.c: In the SIOCGIFMEDIA ioctl in ifmedia_ioctl(), avoid calling malloc() with M_WAITOK. Return an error if the M_NOWAIT malloc fails. The ti(4) driver and the wi(4) driver, at least, call this with a mutex held. This causes witness warnings for 'ifconfig -a' with a wi(4) or ti(4) board in the system. (I've only verified for ti(4)). ip_output.c: Fragment large datagrams so that each segment contains a multiple of PAGE_SIZE amount of data plus headers. This allows the receiver to potentially do page flipping on receives. if_ti.c: Add zero copy receive support to the ti(4) driver. If TI_PRIVATE_JUMBOS is not defined, it now uses the jumbo(9) buffer allocator for jumbo receive buffers. Add a new character device interface for the ti(4) driver for the new debugging interface. This allows (a patched version of) gdb to talk to the Tigon board and debug the firmware. There are also a few additional debugging ioctls available through this interface. Add header splitting support to the ti(4) driver. Tweak some of the default interrupt coalescing parameters to more useful defaults. Add hooks for supporting transmit flow control, but leave it turned off with a comment describing why it is turned off. if_tireg.h: Change the firmware rev to 12.4.11, since we're really at 12.4.11 plus fixes from 12.4.13. Add defines needed for debugging. Remove the ti_stats structure, it is now defined in sys/tiio.h. ti_fw.h: 12.4.11 firmware. ti_fw2.h: 12.4.11 firmware, plus selected fixes from 12.4.13, and my header splitting patches. Revision 12.4.13 doesn't handle 10/100 negotiation properly. (This firmware is the same as what was in the tree previously, with the addition of header splitting support.) sys/jumbo.h: Jumbo buffer allocator interface. sys/mbuf.h: Add a new external mbuf type, EXT_DISPOSABLE, to indicate that the payload buffer can be thrown away / flipped to a userland process. socketvar.h: Add prototype for socow_setup. tiio.h: ioctl interface to the character portion of the ti(4) driver, plus associated structure/type definitions. uio.h: Change prototype for uiomoveco() so that we'll know whether the source page is disposable. ufs_readwrite.c:Update for new prototype of uiomoveco(). vm_fault.c: In vm_fault(), check to see whether we need to do a page based copy on write fault. vm_object.c: Add a new function, vm_object_allocate_wait(). This does the same thing that vm_object allocate does, except that it gives the caller the opportunity to specify whether it should wait on the uma_zalloc() of the object structre. This allows vm objects to be allocated while holding a mutex. (Without generating WITNESS warnings.) vm_object_allocate() is implemented as a call to vm_object_allocate_wait() with the malloc flag set to M_WAITOK. vm_object.h: Add prototype for vm_object_allocate_wait(). vm_page.c: Add page-based copy on write setup, clear and fault routines. vm_page.h: Add page based COW function prototypes and variable in the vm_page structure. Many thanks to Drew Gallatin, who wrote the zero copy send and receive code, and to all the other folks who have tested and reviewed this code over the years.
2002-06-26 03:37:47 +00:00
/*
* Maximum register address on the Tigon.
*/
#define TI_REG_MAX 0x3fff
/*
* These values were taken from Alteon's tg.h.
*/
#define TI_BEG_SRAM 0x0 /* host thinks it's here */
#define TI_BEG_SCRATCH 0xc00000 /* beg of scratch pad area */
#define TI_END_SRAM_II 0x800000 /* end of SRAM, for 2 MB stuffed */
#define TI_END_SCRATCH_II 0xc04000 /* end of scratch pad CPU A (16KB) */
#define TI_END_SCRATCH_B 0xc02000 /* end of scratch pad CPU B (8KB) */
#define TI_BEG_SCRATCH_B_DEBUG 0xd00000 /* beg of scratch pad for ioctl */
#define TI_END_SCRATCH_B_DEBUG 0xd02000 /* end of scratch pad for ioctl */
#define TI_SCRATCH_DEBUG_OFF 0x100000 /* offset for ioctl usage */
#define TI_END_SRAM_I 0x200000 /* end of SRAM, for 2 MB stuffed */
#define TI_END_SCRATCH_I 0xc00800 /* end of scratch pad area (2KB) */
#define TI_BEG_PROM 0x40000000 /* beg of PROM, special access */
#define TI_BEG_FLASH 0x80000000 /* beg of EEPROM, special access */
#define TI_END_FLASH 0x80100000 /* end of EEPROM for 1 MB stuff */
#define TI_BEG_SER_EEPROM 0xa0000000 /* beg of Serial EEPROM (fake out) */
#define TI_END_SER_EEPROM 0xa0002000 /* end of Serial EEPROM (fake out) */
#define TI_BEG_REGS 0xc0000000 /* beg of register area */
#define TI_END_REGS 0xc0000400 /* end of register area */
#define TI_END_WRITE_REGS 0xc0000180 /* can't write GPRs currently */
#define TI_BEG_REGS2 0xc0000200 /* beg of second writeable reg area */
/* the EEPROM is byte addressable in a pretty odd way */
#define EEPROM_BYTE_LOC 0xff000000
/*
* From Alteon's tg.h.
*/
#define TI_PROCESSOR_A 0
#define TI_PROCESSOR_B 1
#define TI_CPU_A TG_PROCESSOR_A
#define TI_CPU_B TG_PROCESSOR_B
/*
* Following macro can be used to access to any of the CPU registers
* It will adjust the address appropriately.
* Parameters:
* reg - The register to access, e.g TI_CPU_CONTROL
* cpu - cpu, i.e PROCESSOR_A or PROCESSOR_B (or TI_CPU_A or TI_CPU_B)
*/
#define CPU_REG(reg, cpu) ((reg) + (cpu) * 0x100)
/*
* Even on the alpha, pci addresses are 32-bit quantities
*/
typedef struct {
2011-11-04 19:12:07 +00:00
uint32_t ti_addr_hi;
uint32_t ti_addr_lo;
} ti_hostaddr;
#define TI_HOSTADDR(x) x.ti_addr_lo
static __inline void
ti_hostaddr64(ti_hostaddr *x, bus_addr_t addr)
{
uint64_t baddr;
baddr = (uint64_t)addr;
x->ti_addr_lo = baddr & 0xffffffff;
x->ti_addr_hi = baddr >> 32;
}
/*
* Ring control block structure. The rules for the max_len field
* are as follows:
*
* For the send ring, max_len indicates the number of entries in the
* ring (128, 256 or 512).
*
* For the standard receive ring, max_len indicates the threshold
* used to decide when a frame should be put in the jumbo receive ring
* instead of the standard one.
*
* For the mini ring, max_len indicates the size of the buffers in the
* ring. This is the value used to decide when a frame is small enough
* to be placed in the mini ring.
*
* For the return receive ring, max_len indicates the number of entries
* in the ring. It can be one of 2048, 1024 or 0 (which is the same as
* 2048 for backwards compatibility). The value 1024 can only be used
* if the mini ring is disabled.
*/
struct ti_rcb {
ti_hostaddr ti_hostaddr;
#if BYTE_ORDER == BIG_ENDIAN
2011-11-04 19:12:07 +00:00
uint16_t ti_max_len;
uint16_t ti_flags;
#else
2011-11-04 19:12:07 +00:00
uint16_t ti_flags;
uint16_t ti_max_len;
#endif
2011-11-04 19:12:07 +00:00
uint32_t ti_unused;
};
#define TI_RCB_FLAG_TCP_UDP_CKSUM 0x00000001
#define TI_RCB_FLAG_IP_CKSUM 0x00000002
#define TI_RCB_FLAG_NO_PHDR_CKSUM 0x00000008
#define TI_RCB_FLAG_VLAN_ASSIST 0x00000010
#define TI_RCB_FLAG_COAL_UPD_ONLY 0x00000020
#define TI_RCB_FLAG_HOST_RING 0x00000040
#define TI_RCB_FLAG_IEEE_SNAP_CKSUM 0x00000080
#define TI_RCB_FLAG_USE_EXT_RX_BD 0x00000100
#define TI_RCB_FLAG_RING_DISABLED 0x00000200
struct ti_producer {
2011-11-04 19:12:07 +00:00
uint32_t ti_idx;
uint32_t ti_unused;
};
/*
* Tigon general information block. This resides in host memory
* and contains the status counters, ring control blocks and
* producer pointers.
*/
struct ti_gib {
struct ti_stats ti_stats;
struct ti_rcb ti_ev_rcb;
struct ti_rcb ti_cmd_rcb;
struct ti_rcb ti_tx_rcb;
struct ti_rcb ti_std_rx_rcb;
struct ti_rcb ti_jumbo_rx_rcb;
struct ti_rcb ti_mini_rx_rcb;
struct ti_rcb ti_return_rcb;
ti_hostaddr ti_ev_prodidx_ptr;
ti_hostaddr ti_return_prodidx_ptr;
ti_hostaddr ti_tx_considx_ptr;
ti_hostaddr ti_refresh_stats_ptr;
};
/*
* Buffer descriptor structures. There are basically three types
* of structures: normal receive descriptors, extended receive
* descriptors and transmit descriptors. The extended receive
* descriptors are optionally used only for the jumbo receive ring.
*/
struct ti_rx_desc {
ti_hostaddr ti_addr;
#if BYTE_ORDER == BIG_ENDIAN
2011-11-04 19:12:07 +00:00
uint16_t ti_idx;
uint16_t ti_len;
#else
2011-11-04 19:12:07 +00:00
uint16_t ti_len;
uint16_t ti_idx;
#endif
#if BYTE_ORDER == BIG_ENDIAN
2011-11-04 19:12:07 +00:00
uint16_t ti_type;
uint16_t ti_flags;
#else
2011-11-04 19:12:07 +00:00
uint16_t ti_flags;
uint16_t ti_type;
#endif
#if BYTE_ORDER == BIG_ENDIAN
2011-11-04 19:12:07 +00:00
uint16_t ti_ip_cksum;
uint16_t ti_tcp_udp_cksum;
#else
2011-11-04 19:12:07 +00:00
uint16_t ti_tcp_udp_cksum;
uint16_t ti_ip_cksum;
#endif
#if BYTE_ORDER == BIG_ENDIAN
2011-11-04 19:12:07 +00:00
uint16_t ti_error_flags;
uint16_t ti_vlan_tag;
#else
2011-11-04 19:12:07 +00:00
uint16_t ti_vlan_tag;
uint16_t ti_error_flags;
#endif
2011-11-04 19:12:07 +00:00
uint32_t ti_rsvd;
uint32_t ti_opaque;
};
Overhaul bus_dma(9) usage in driver: - Don't use a single big DMA block for all rings. Create separate DMA area for each ring instead. Currently the following DMA areas are created: Event ring, standard RX ring, jumbo RX ring, RX return ring, hardware MAC statistics and producer/consumer status area. For Tigon II, mini RX ring and TX ring are additionally created. - Added missing bus_dmamap_sync(9) in various TX/RX paths. - TX ring is no longer created for Tigon 1 such that it saves more resources on Tigon 1. - Data sheet is not clear about alignment requirement of each ring so use 32 bytes alignment for normal DMA area but use 64 bytes alignment for jumbo RX ring where the extended RX descriptor size is 64 bytes. - For each TX/RX buffers use separate DMA tag(e.g. the size of a DMA segment, total size of DMA segments etc). - Tigon allows separate DMA area for event producer, RX return producer and TX consumer which is really cool feature. This means TX and RX path could be independently run in parallel. However ti(4) uses a single driver lock so it's meaningless to have separate DMA area for these producer/consumer such that this change creates a single status DMA area. - It seems Tigon has no limits on DMA address space and I also don't see any problem with that but old comments in driver indicates there could be issues on descriptors being located in 64bit region. Introduce a tunable, dev.ti.%d.dac, to disable using 64bit DMA in driver. The default is 0 which means it would use full 64bit DMA. If there are DMA issues, users can disable it by setting the tunable to 0. - Do not increase watchdog timer in ti_txeof(). Previously driver increased the watchdog timer whenever there are queued TX frames. - When stat ticks is set to 0, skip processing ti_stats_update(), avoiding bus_dmamap_sync(9) and updating if_collisions counter. - MTU does not include FCS bytes, replace it with ETHER_VLAN_ENCAP_LEN. With these changes, ti(4) should work on PAE environments. Many thanks to Jay Borkenhagen for remote hardware access.
2011-11-14 20:38:14 +00:00
#define TI_STD_RX_RING_SZ (sizeof(struct ti_rx_desc) * TI_STD_RX_RING_CNT)
#define TI_MINI_RX_RING_SZ (sizeof(struct ti_rx_desc) * TI_MINI_RX_RING_CNT)
#define TI_RX_RETURN_RING_SZ (sizeof(struct ti_rx_desc) * TI_RETURN_RING_CNT)
struct ti_rx_desc_ext {
ti_hostaddr ti_addr1;
ti_hostaddr ti_addr2;
ti_hostaddr ti_addr3;
#if BYTE_ORDER == BIG_ENDIAN
2011-11-04 19:12:07 +00:00
uint16_t ti_len1;
uint16_t ti_len2;
#else
2011-11-04 19:12:07 +00:00
uint16_t ti_len2;
uint16_t ti_len1;
#endif
#if BYTE_ORDER == BIG_ENDIAN
2011-11-04 19:12:07 +00:00
uint16_t ti_len3;
uint16_t ti_rsvd0;
#else
2011-11-04 19:12:07 +00:00
uint16_t ti_rsvd0;
uint16_t ti_len3;
#endif
ti_hostaddr ti_addr0;
#if BYTE_ORDER == BIG_ENDIAN
2011-11-04 19:12:07 +00:00
uint16_t ti_idx;
uint16_t ti_len0;
#else
2011-11-04 19:12:07 +00:00
uint16_t ti_len0;
uint16_t ti_idx;
#endif
#if BYTE_ORDER == BIG_ENDIAN
2011-11-04 19:12:07 +00:00
uint16_t ti_type;
uint16_t ti_flags;
#else
2011-11-04 19:12:07 +00:00
uint16_t ti_flags;
uint16_t ti_type;
#endif
#if BYTE_ORDER == BIG_ENDIAN
2011-11-04 19:12:07 +00:00
uint16_t ti_ip_cksum;
uint16_t ti_tcp_udp_cksum;
#else
2011-11-04 19:12:07 +00:00
uint16_t ti_tcp_udp_cksum;
uint16_t ti_ip_cksum;
#endif
#if BYTE_ORDER == BIG_ENDIAN
2011-11-04 19:12:07 +00:00
uint16_t ti_error_flags;
uint16_t ti_vlan_tag;
#else
2011-11-04 19:12:07 +00:00
uint16_t ti_vlan_tag;
uint16_t ti_error_flags;
#endif
2011-11-04 19:12:07 +00:00
uint32_t ti_rsvd1;
uint32_t ti_opaque;
};
Overhaul bus_dma(9) usage in driver: - Don't use a single big DMA block for all rings. Create separate DMA area for each ring instead. Currently the following DMA areas are created: Event ring, standard RX ring, jumbo RX ring, RX return ring, hardware MAC statistics and producer/consumer status area. For Tigon II, mini RX ring and TX ring are additionally created. - Added missing bus_dmamap_sync(9) in various TX/RX paths. - TX ring is no longer created for Tigon 1 such that it saves more resources on Tigon 1. - Data sheet is not clear about alignment requirement of each ring so use 32 bytes alignment for normal DMA area but use 64 bytes alignment for jumbo RX ring where the extended RX descriptor size is 64 bytes. - For each TX/RX buffers use separate DMA tag(e.g. the size of a DMA segment, total size of DMA segments etc). - Tigon allows separate DMA area for event producer, RX return producer and TX consumer which is really cool feature. This means TX and RX path could be independently run in parallel. However ti(4) uses a single driver lock so it's meaningless to have separate DMA area for these producer/consumer such that this change creates a single status DMA area. - It seems Tigon has no limits on DMA address space and I also don't see any problem with that but old comments in driver indicates there could be issues on descriptors being located in 64bit region. Introduce a tunable, dev.ti.%d.dac, to disable using 64bit DMA in driver. The default is 0 which means it would use full 64bit DMA. If there are DMA issues, users can disable it by setting the tunable to 0. - Do not increase watchdog timer in ti_txeof(). Previously driver increased the watchdog timer whenever there are queued TX frames. - When stat ticks is set to 0, skip processing ti_stats_update(), avoiding bus_dmamap_sync(9) and updating if_collisions counter. - MTU does not include FCS bytes, replace it with ETHER_VLAN_ENCAP_LEN. With these changes, ti(4) should work on PAE environments. Many thanks to Jay Borkenhagen for remote hardware access.
2011-11-14 20:38:14 +00:00
#ifdef TI_SF_BUF_JUMBO
#define TI_JUMBO_RX_RING_SZ \
(sizeof(struct ti_rx_desc_ext) * TI_JUMBO_RX_RING_CNT)
#else
#define TI_JUMBO_RX_RING_SZ \
(sizeof(struct ti_rx_desc) * TI_JUMBO_RX_RING_CNT)
#endif
/*
* Transmit descriptors are, mercifully, very small.
*/
struct ti_tx_desc {
ti_hostaddr ti_addr;
#if BYTE_ORDER == BIG_ENDIAN
2011-11-04 19:12:07 +00:00
uint16_t ti_len;
uint16_t ti_flags;
#else
2011-11-04 19:12:07 +00:00
uint16_t ti_flags;
uint16_t ti_len;
#endif
#if BYTE_ORDER == BIG_ENDIAN
2011-11-04 19:12:07 +00:00
uint16_t ti_rsvd;
uint16_t ti_vlan_tag;
#else
2011-11-04 19:12:07 +00:00
uint16_t ti_vlan_tag;
uint16_t ti_rsvd;
#endif
};
Overhaul bus_dma(9) usage in driver: - Don't use a single big DMA block for all rings. Create separate DMA area for each ring instead. Currently the following DMA areas are created: Event ring, standard RX ring, jumbo RX ring, RX return ring, hardware MAC statistics and producer/consumer status area. For Tigon II, mini RX ring and TX ring are additionally created. - Added missing bus_dmamap_sync(9) in various TX/RX paths. - TX ring is no longer created for Tigon 1 such that it saves more resources on Tigon 1. - Data sheet is not clear about alignment requirement of each ring so use 32 bytes alignment for normal DMA area but use 64 bytes alignment for jumbo RX ring where the extended RX descriptor size is 64 bytes. - For each TX/RX buffers use separate DMA tag(e.g. the size of a DMA segment, total size of DMA segments etc). - Tigon allows separate DMA area for event producer, RX return producer and TX consumer which is really cool feature. This means TX and RX path could be independently run in parallel. However ti(4) uses a single driver lock so it's meaningless to have separate DMA area for these producer/consumer such that this change creates a single status DMA area. - It seems Tigon has no limits on DMA address space and I also don't see any problem with that but old comments in driver indicates there could be issues on descriptors being located in 64bit region. Introduce a tunable, dev.ti.%d.dac, to disable using 64bit DMA in driver. The default is 0 which means it would use full 64bit DMA. If there are DMA issues, users can disable it by setting the tunable to 0. - Do not increase watchdog timer in ti_txeof(). Previously driver increased the watchdog timer whenever there are queued TX frames. - When stat ticks is set to 0, skip processing ti_stats_update(), avoiding bus_dmamap_sync(9) and updating if_collisions counter. - MTU does not include FCS bytes, replace it with ETHER_VLAN_ENCAP_LEN. With these changes, ti(4) should work on PAE environments. Many thanks to Jay Borkenhagen for remote hardware access.
2011-11-14 20:38:14 +00:00
#define TI_TX_RING_SZ (sizeof(struct ti_tx_desc) * TI_TX_RING_CNT)
/*
* NOTE! On the Alpha, we have an alignment constraint.
* The first thing in the packet is a 14-byte Ethernet header.
* This means that the packet is misaligned. To compensate,
* we actually offset the data 2 bytes into the cluster. This
2005-09-29 15:04:58 +00:00
* aligns the packet after the Ethernet header at a 32-bit
* boundary.
*/
#define TI_FRAMELEN 1518
#define TI_JUMBO_FRAMELEN 9018
#define TI_JUMBO_MTU (TI_JUMBO_FRAMELEN-ETHER_HDR_LEN-ETHER_CRC_LEN)
#define TI_PAGE_SIZE PAGE_SIZE
#define TI_MIN_FRAMELEN 60
/*
* Buffer descriptor error flags.
*/
#define TI_BDERR_CRC 0x0001
#define TI_BDERR_COLLDETECT 0x0002
#define TI_BDERR_LINKLOST 0x0004
#define TI_BDERR_DECODE 0x0008
#define TI_BDERR_ODD_NIBBLES 0x0010
#define TI_BDERR_MAC_ABRT 0x0020
#define TI_BDERR_RUNT 0x0040
#define TI_BDERR_TRUNC 0x0080
#define TI_BDERR_GIANT 0x0100
/*
* Buffer descriptor flags.
*/
#define TI_BDFLAG_TCP_UDP_CKSUM 0x0001
#define TI_BDFLAG_IP_CKSUM 0x0002
#define TI_BDFLAG_END 0x0004
#define TI_BDFLAG_MORE 0x0008
#define TI_BDFLAG_JUMBO_RING 0x0010
#define TI_BDFLAG_UCAST_PKT 0x0020
#define TI_BDFLAG_MCAST_PKT 0x0040
#define TI_BDFLAG_BCAST_PKT 0x0060
#define TI_BDFLAG_IP_FRAG 0x0080
#define TI_BDFLAG_IP_FRAG_END 0x0100
#define TI_BDFLAG_VLAN_TAG 0x0200
#define TI_BDFLAG_ERROR 0x0400
#define TI_BDFLAG_COAL_NOW 0x0800
#define TI_BDFLAG_MINI_RING 0x1000
/*
* Descriptor type flags. I think these only have meaning for
* the Tigon 1. I had to extract them from the sample driver source
* since they aren't in the manual.
*/
#define TI_BDTYPE_TYPE_NULL 0x0000
#define TI_BDTYPE_SEND_BD 0x0001
#define TI_BDTYPE_RECV_BD 0x0002
#define TI_BDTYPE_RECV_JUMBO_BD 0x0003
#define TI_BDTYPE_RECV_BD_LAST 0x0004
#define TI_BDTYPE_SEND_DATA 0x0005
#define TI_BDTYPE_SEND_DATA_LAST 0x0006
#define TI_BDTYPE_RECV_DATA 0x0007
#define TI_BDTYPE_RECV_DATA_LAST 0x000b
#define TI_BDTYPE_EVENT_RUPT 0x000c
#define TI_BDTYPE_EVENT_NO_RUPT 0x000d
#define TI_BDTYPE_ODD_START 0x000e
#define TI_BDTYPE_UPDATE_STATS 0x000f
#define TI_BDTYPE_SEND_DUMMY_DMA 0x0010
#define TI_BDTYPE_EVENT_PROD 0x0011
#define TI_BDTYPE_TX_CONS 0x0012
#define TI_BDTYPE_RX_PROD 0x0013
#define TI_BDTYPE_REFRESH_STATS 0x0014
#define TI_BDTYPE_SEND_DATA_LAST_VLAN 0x0015
#define TI_BDTYPE_SEND_DATA_COAL 0x0016
#define TI_BDTYPE_SEND_DATA_LAST_COAL 0x0017
#define TI_BDTYPE_SEND_DATA_LAST_VLAN_COAL 0x0018
#define TI_BDTYPE_TX_CONS_NO_INTR 0x0019
/*
* Tigon command structure.
*/
struct ti_cmd_desc {
2011-11-04 19:12:07 +00:00
uint32_t ti_cmdx;
};
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
#define TI_CMD_CMD(cmd) (((((cmd)->ti_cmdx)) >> 24) & 0xff)
#define TI_CMD_CODE(cmd) (((((cmd)->ti_cmdx)) >> 12) & 0xfff)
#define TI_CMD_IDX(cmd) ((((cmd)->ti_cmdx)) & 0xfff)
#define TI_CMD_HOST_STATE 0x01
#define TI_CMD_CODE_STACK_UP 0x01
#define TI_CMD_CODE_STACK_DOWN 0x02
/*
* This command enables software address filtering. It's a workaround
* for a bug in the Tigon 1 and not implemented for the Tigon 2.
*/
#define TI_CMD_FDR_FILTERING 0x02
#define TI_CMD_CODE_FILT_ENB 0x01
#define TI_CMD_CODE_FILT_DIS 0x02
#define TI_CMD_SET_RX_PROD_IDX 0x03 /* obsolete */
#define TI_CMD_UPDATE_GENCOM 0x04
#define TI_CMD_RESET_JUMBO_RING 0x05
#define TI_CMD_SET_PARTIAL_RX_CNT 0x06
#define TI_CMD_ADD_MCAST_ADDR 0x08 /* obsolete */
#define TI_CMD_DEL_MCAST_ADDR 0x09 /* obsolete */
#define TI_CMD_SET_PROMISC_MODE 0x0A
#define TI_CMD_CODE_PROMISC_ENB 0x01
#define TI_CMD_CODE_PROMISC_DIS 0x02
#define TI_CMD_LINK_NEGOTIATION 0x0B
#define TI_CMD_CODE_NEGOTIATE_BOTH 0x00
#define TI_CMD_CODE_NEGOTIATE_GIGABIT 0x01
#define TI_CMD_CODE_NEGOTIATE_10_100 0x02
#define TI_CMD_SET_MAC_ADDR 0x0C
#define TI_CMD_CLR_PROFILE 0x0D
#define TI_CMD_SET_ALLMULTI 0x0E
#define TI_CMD_CODE_ALLMULTI_ENB 0x01
#define TI_CMD_CODE_ALLMULTI_DIS 0x02
#define TI_CMD_CLR_STATS 0x0F
#define TI_CMD_SET_RX_JUMBO_PROD_IDX 0x10 /* obsolete */
#define TI_CMD_RFRSH_STATS 0x11
#define TI_CMD_EXT_ADD_MCAST 0x12
#define TI_CMD_EXT_DEL_MCAST 0x13
/*
* Utility macros to make issuing commands a little simpler. Assumes
* that 'sc' and 'cmd' are in local scope.
*/
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
#define TI_DO_CMD(x, y, z) do { \
cmd.ti_cmdx = (((x) << 24) | ((y) << 12) | ((z))); \
ti_cmd(sc, &cmd); \
} while(0)
#define TI_DO_CMD_EXT(x, y, z, v, w) do { \
cmd.ti_cmdx = (((x) << 24) | ((y) << 12) | ((z))); \
ti_cmd_ext(sc, &cmd, (v), (w)); \
} while(0)
/*
* Other utility macros.
*/
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
#define TI_INC(x, y) (x) = ((x) + 1) % y
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
#define TI_UPDATE_JUMBOPROD(x, y) do { \
if ((x)->ti_hwrev == TI_HWREV_TIGON) \
TI_DO_CMD(TI_CMD_SET_RX_JUMBO_PROD_IDX, 0, (y)); \
else \
CSR_WRITE_4((x), TI_MB_JUMBORXPROD_IDX, (y)); \
} while(0)
#define TI_UPDATE_MINIPROD(x, y) \
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
CSR_WRITE_4((x), TI_MB_MINIRXPROD_IDX, (y))
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
#define TI_UPDATE_STDPROD(x, y) do { \
if ((x)->ti_hwrev == TI_HWREV_TIGON) \
TI_DO_CMD(TI_CMD_SET_RX_PROD_IDX, 0, (y)); \
else \
CSR_WRITE_4((x), TI_MB_STDRXPROD_IDX, (y)); \
} while(0)
/*
* Tigon event structure.
*/
struct ti_event_desc {
2011-11-04 19:12:07 +00:00
uint32_t ti_eventx;
uint32_t ti_rsvd;
};
Overhaul bus_dma(9) usage in driver: - Don't use a single big DMA block for all rings. Create separate DMA area for each ring instead. Currently the following DMA areas are created: Event ring, standard RX ring, jumbo RX ring, RX return ring, hardware MAC statistics and producer/consumer status area. For Tigon II, mini RX ring and TX ring are additionally created. - Added missing bus_dmamap_sync(9) in various TX/RX paths. - TX ring is no longer created for Tigon 1 such that it saves more resources on Tigon 1. - Data sheet is not clear about alignment requirement of each ring so use 32 bytes alignment for normal DMA area but use 64 bytes alignment for jumbo RX ring where the extended RX descriptor size is 64 bytes. - For each TX/RX buffers use separate DMA tag(e.g. the size of a DMA segment, total size of DMA segments etc). - Tigon allows separate DMA area for event producer, RX return producer and TX consumer which is really cool feature. This means TX and RX path could be independently run in parallel. However ti(4) uses a single driver lock so it's meaningless to have separate DMA area for these producer/consumer such that this change creates a single status DMA area. - It seems Tigon has no limits on DMA address space and I also don't see any problem with that but old comments in driver indicates there could be issues on descriptors being located in 64bit region. Introduce a tunable, dev.ti.%d.dac, to disable using 64bit DMA in driver. The default is 0 which means it would use full 64bit DMA. If there are DMA issues, users can disable it by setting the tunable to 0. - Do not increase watchdog timer in ti_txeof(). Previously driver increased the watchdog timer whenever there are queued TX frames. - When stat ticks is set to 0, skip processing ti_stats_update(), avoiding bus_dmamap_sync(9) and updating if_collisions counter. - MTU does not include FCS bytes, replace it with ETHER_VLAN_ENCAP_LEN. With these changes, ti(4) should work on PAE environments. Many thanks to Jay Borkenhagen for remote hardware access.
2011-11-14 20:38:14 +00:00
#define TI_EVENT_RING_SZ (sizeof(struct ti_event_desc) * TI_EVENT_RING_CNT)
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
#define TI_EVENT_EVENT(e) (((((e)->ti_eventx)) >> 24) & 0xff)
#define TI_EVENT_CODE(e) (((((e)->ti_eventx)) >> 12) & 0xfff)
#define TI_EVENT_IDX(e) (((((e)->ti_eventx))) & 0xfff)
/*
* Tigon events.
*/
#define TI_EV_FIRMWARE_UP 0x01
#define TI_EV_STATS_UPDATED 0x04
#define TI_EV_LINKSTAT_CHANGED 0x06
#define TI_EV_CODE_GIG_LINK_UP 0x01
#define TI_EV_CODE_LINK_DOWN 0x02
#define TI_EV_CODE_LINK_UP 0x03
#define TI_EV_ERROR 0x07
#define TI_EV_CODE_ERR_INVAL_CMD 0x01
#define TI_EV_CODE_ERR_UNIMP_CMD 0x02
#define TI_EV_CODE_ERR_BADCFG 0x03
#define TI_EV_MCAST_UPDATED 0x08
#define TI_EV_CODE_MCAST_ADD 0x01
#define TI_EV_CODE_MCAST_DEL 0x02
#define TI_EV_RESET_JUMBO_RING 0x09
/*
* Register access macros. The Tigon always uses memory mapped register
* accesses and all registers must be accessed with 32 bit operations.
*/
#define CSR_WRITE_4(sc, reg, val) \
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
bus_space_write_4((sc)->ti_btag, (sc)->ti_bhandle, (reg), (val))
#define CSR_READ_4(sc, reg) \
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
bus_space_read_4((sc)->ti_btag, (sc)->ti_bhandle, (reg))
#define TI_SETBIT(sc, reg, x) \
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
CSR_WRITE_4((sc), (reg), (CSR_READ_4((sc), (reg)) | (x)))
#define TI_CLRBIT(sc, reg, x) \
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
CSR_WRITE_4((sc), (reg), (CSR_READ_4((sc), (reg)) & ~(x)))
- Tx side bus_dmamap_load_mbuf_sg(9) support. This reduces bookkeeping requiried to keep consistent softc state before/after callback function invocation and supposed to be sligntly faster than previous one as it wouldn't incur callback overhead. With this change callback function was gone. - Decrease TI_MAXTXSEGS to 32 from 128. It seems that most mbuf chain length is less than 32 and it would be re-packed with m_defrag(9) if its chain length is larger than TI_MAXTXSEGS. This would protect ti(4) against possible kernel stack overflow when txsegs[] is put on stack. Alternatively, we can embed the txsegs[] into softc. However, that would waste memory and make Tx/Rx speration hard when we want to sperate Tx/Rx handlers to optimize locking. - Fix dma map tracking used in Tx path. Previously it used the dma map of the last mbuf chain in ti_txeof() which was incorrect as ti(4) used dma map of the first mbuf chain when it loads a mbuf chain with bus_dmamap_load_mbuf(9). Correct the bug by introducing queues that keep track of active/inactive dma maps/mbuf chain. - Use ti_txcnt to check whether driver need to set watchdog timer instead of blidnly clearing the timer in ti_txeof(). - Remove the 3rd arg. of ti_encap(). Since ti(4) now caches the last descriptor index(ti_tx_saved_prodidx) used in Tx there is no need to pass it as a fuction arg. - Change data type of producer/consumer index to int from u_int16_t in order to remove implicit type conversions in Tx/Rx handlers. - Check interface queue before getting a mbuf chain to reduce locking overhead. - Check number of available Tx descriptores to be 16 or higher in ti_start(). This wouldn't protect Tx descriptor shortage but it would reduce number of bus_dmamap_unload(9) calls in ti_encap() when we are about to running out of Tx descriptors. - Command NIC to send packets ony when the driver really has packets enqueued. Previously it always set TI_MB_SENDPROD_IDX which would command NIC to DMA Tx descriptors into NIC local memory regardless of Tx descriptor changes. Reviewed by: scottl
2006-01-03 06:14:07 +00:00
struct ti_txdesc {
struct mbuf *tx_m;
bus_dmamap_t tx_dmamap;
STAILQ_ENTRY(ti_txdesc) tx_q;
};
STAILQ_HEAD(ti_txdq, ti_txdesc);
Overhaul bus_dma(9) usage in driver: - Don't use a single big DMA block for all rings. Create separate DMA area for each ring instead. Currently the following DMA areas are created: Event ring, standard RX ring, jumbo RX ring, RX return ring, hardware MAC statistics and producer/consumer status area. For Tigon II, mini RX ring and TX ring are additionally created. - Added missing bus_dmamap_sync(9) in various TX/RX paths. - TX ring is no longer created for Tigon 1 such that it saves more resources on Tigon 1. - Data sheet is not clear about alignment requirement of each ring so use 32 bytes alignment for normal DMA area but use 64 bytes alignment for jumbo RX ring where the extended RX descriptor size is 64 bytes. - For each TX/RX buffers use separate DMA tag(e.g. the size of a DMA segment, total size of DMA segments etc). - Tigon allows separate DMA area for event producer, RX return producer and TX consumer which is really cool feature. This means TX and RX path could be independently run in parallel. However ti(4) uses a single driver lock so it's meaningless to have separate DMA area for these producer/consumer such that this change creates a single status DMA area. - It seems Tigon has no limits on DMA address space and I also don't see any problem with that but old comments in driver indicates there could be issues on descriptors being located in 64bit region. Introduce a tunable, dev.ti.%d.dac, to disable using 64bit DMA in driver. The default is 0 which means it would use full 64bit DMA. If there are DMA issues, users can disable it by setting the tunable to 0. - Do not increase watchdog timer in ti_txeof(). Previously driver increased the watchdog timer whenever there are queued TX frames. - When stat ticks is set to 0, skip processing ti_stats_update(), avoiding bus_dmamap_sync(9) and updating if_collisions counter. - MTU does not include FCS bytes, replace it with ETHER_VLAN_ENCAP_LEN. With these changes, ti(4) should work on PAE environments. Many thanks to Jay Borkenhagen for remote hardware access.
2011-11-14 20:38:14 +00:00
struct ti_status {
/*
* Make sure producer structures are aligned on 32-byte cache
Overhaul bus_dma(9) usage in driver: - Don't use a single big DMA block for all rings. Create separate DMA area for each ring instead. Currently the following DMA areas are created: Event ring, standard RX ring, jumbo RX ring, RX return ring, hardware MAC statistics and producer/consumer status area. For Tigon II, mini RX ring and TX ring are additionally created. - Added missing bus_dmamap_sync(9) in various TX/RX paths. - TX ring is no longer created for Tigon 1 such that it saves more resources on Tigon 1. - Data sheet is not clear about alignment requirement of each ring so use 32 bytes alignment for normal DMA area but use 64 bytes alignment for jumbo RX ring where the extended RX descriptor size is 64 bytes. - For each TX/RX buffers use separate DMA tag(e.g. the size of a DMA segment, total size of DMA segments etc). - Tigon allows separate DMA area for event producer, RX return producer and TX consumer which is really cool feature. This means TX and RX path could be independently run in parallel. However ti(4) uses a single driver lock so it's meaningless to have separate DMA area for these producer/consumer such that this change creates a single status DMA area. - It seems Tigon has no limits on DMA address space and I also don't see any problem with that but old comments in driver indicates there could be issues on descriptors being located in 64bit region. Introduce a tunable, dev.ti.%d.dac, to disable using 64bit DMA in driver. The default is 0 which means it would use full 64bit DMA. If there are DMA issues, users can disable it by setting the tunable to 0. - Do not increase watchdog timer in ti_txeof(). Previously driver increased the watchdog timer whenever there are queued TX frames. - When stat ticks is set to 0, skip processing ti_stats_update(), avoiding bus_dmamap_sync(9) and updating if_collisions counter. - MTU does not include FCS bytes, replace it with ETHER_VLAN_ENCAP_LEN. With these changes, ti(4) should work on PAE environments. Many thanks to Jay Borkenhagen for remote hardware access.
2011-11-14 20:38:14 +00:00
* line boundaries. We can create separate DMA area for each
* producer/consumer area but it wouldn't get much benefit
* since driver use a global driver lock.
*/
struct ti_producer ti_ev_prodidx_r;
2011-11-04 19:12:07 +00:00
uint32_t ti_pad0[6];
struct ti_producer ti_return_prodidx_r;
2011-11-04 19:12:07 +00:00
uint32_t ti_pad1[6];
struct ti_producer ti_tx_considx_r;
2011-11-04 19:12:07 +00:00
uint32_t ti_pad2[6];
};
Overhaul bus_dma(9) usage in driver: - Don't use a single big DMA block for all rings. Create separate DMA area for each ring instead. Currently the following DMA areas are created: Event ring, standard RX ring, jumbo RX ring, RX return ring, hardware MAC statistics and producer/consumer status area. For Tigon II, mini RX ring and TX ring are additionally created. - Added missing bus_dmamap_sync(9) in various TX/RX paths. - TX ring is no longer created for Tigon 1 such that it saves more resources on Tigon 1. - Data sheet is not clear about alignment requirement of each ring so use 32 bytes alignment for normal DMA area but use 64 bytes alignment for jumbo RX ring where the extended RX descriptor size is 64 bytes. - For each TX/RX buffers use separate DMA tag(e.g. the size of a DMA segment, total size of DMA segments etc). - Tigon allows separate DMA area for event producer, RX return producer and TX consumer which is really cool feature. This means TX and RX path could be independently run in parallel. However ti(4) uses a single driver lock so it's meaningless to have separate DMA area for these producer/consumer such that this change creates a single status DMA area. - It seems Tigon has no limits on DMA address space and I also don't see any problem with that but old comments in driver indicates there could be issues on descriptors being located in 64bit region. Introduce a tunable, dev.ti.%d.dac, to disable using 64bit DMA in driver. The default is 0 which means it would use full 64bit DMA. If there are DMA issues, users can disable it by setting the tunable to 0. - Do not increase watchdog timer in ti_txeof(). Previously driver increased the watchdog timer whenever there are queued TX frames. - When stat ticks is set to 0, skip processing ti_stats_update(), avoiding bus_dmamap_sync(9) and updating if_collisions counter. - MTU does not include FCS bytes, replace it with ETHER_VLAN_ENCAP_LEN. With these changes, ti(4) should work on PAE environments. Many thanks to Jay Borkenhagen for remote hardware access.
2011-11-14 20:38:14 +00:00
/*
* Ring structures. Most of these reside in host memory and we tell
* the NIC where they are via the ring control blocks. The exceptions
* are the tx and command rings, which live in NIC memory and which
* we access via the shared memory window.
*/
struct ti_ring_data {
struct ti_gib *ti_info;
bus_addr_t ti_info_paddr;
struct ti_status *ti_status;
bus_addr_t ti_status_paddr;
struct ti_rx_desc *ti_rx_std_ring;
bus_addr_t ti_rx_std_ring_paddr;
#ifdef TI_SF_BUF_JUMBO
struct ti_rx_desc_ext *ti_rx_jumbo_ring;
#else
struct ti_rx_desc *ti_rx_jumbo_ring;
#endif
bus_addr_t ti_rx_jumbo_ring_paddr;
struct ti_rx_desc *ti_rx_mini_ring;
bus_addr_t ti_rx_mini_ring_paddr;
struct ti_rx_desc *ti_rx_return_ring;
bus_addr_t ti_rx_return_ring_paddr;
struct ti_event_desc *ti_event_ring;
bus_addr_t ti_event_ring_paddr;
struct ti_tx_desc *ti_tx_ring;
bus_addr_t ti_tx_ring_paddr;
};
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
/*
* Mbuf pointers. We need these to keep track of the virtual addresses
* of our mbuf chains since we can only convert from physical to virtual,
* not the other way around.
*/
struct ti_chain_data {
Overhaul bus_dma(9) usage in driver: - Don't use a single big DMA block for all rings. Create separate DMA area for each ring instead. Currently the following DMA areas are created: Event ring, standard RX ring, jumbo RX ring, RX return ring, hardware MAC statistics and producer/consumer status area. For Tigon II, mini RX ring and TX ring are additionally created. - Added missing bus_dmamap_sync(9) in various TX/RX paths. - TX ring is no longer created for Tigon 1 such that it saves more resources on Tigon 1. - Data sheet is not clear about alignment requirement of each ring so use 32 bytes alignment for normal DMA area but use 64 bytes alignment for jumbo RX ring where the extended RX descriptor size is 64 bytes. - For each TX/RX buffers use separate DMA tag(e.g. the size of a DMA segment, total size of DMA segments etc). - Tigon allows separate DMA area for event producer, RX return producer and TX consumer which is really cool feature. This means TX and RX path could be independently run in parallel. However ti(4) uses a single driver lock so it's meaningless to have separate DMA area for these producer/consumer such that this change creates a single status DMA area. - It seems Tigon has no limits on DMA address space and I also don't see any problem with that but old comments in driver indicates there could be issues on descriptors being located in 64bit region. Introduce a tunable, dev.ti.%d.dac, to disable using 64bit DMA in driver. The default is 0 which means it would use full 64bit DMA. If there are DMA issues, users can disable it by setting the tunable to 0. - Do not increase watchdog timer in ti_txeof(). Previously driver increased the watchdog timer whenever there are queued TX frames. - When stat ticks is set to 0, skip processing ti_stats_update(), avoiding bus_dmamap_sync(9) and updating if_collisions counter. - MTU does not include FCS bytes, replace it with ETHER_VLAN_ENCAP_LEN. With these changes, ti(4) should work on PAE environments. Many thanks to Jay Borkenhagen for remote hardware access.
2011-11-14 20:38:14 +00:00
bus_dma_tag_t ti_parent_tag;
bus_dma_tag_t ti_gib_tag;
bus_dmamap_t ti_gib_map;
bus_dma_tag_t ti_event_ring_tag;
bus_dmamap_t ti_event_ring_map;
bus_dma_tag_t ti_status_tag;
bus_dmamap_t ti_status_map;
bus_dma_tag_t ti_tx_ring_tag;
bus_dmamap_t ti_tx_ring_map;
bus_dma_tag_t ti_tx_tag;
- Tx side bus_dmamap_load_mbuf_sg(9) support. This reduces bookkeeping requiried to keep consistent softc state before/after callback function invocation and supposed to be sligntly faster than previous one as it wouldn't incur callback overhead. With this change callback function was gone. - Decrease TI_MAXTXSEGS to 32 from 128. It seems that most mbuf chain length is less than 32 and it would be re-packed with m_defrag(9) if its chain length is larger than TI_MAXTXSEGS. This would protect ti(4) against possible kernel stack overflow when txsegs[] is put on stack. Alternatively, we can embed the txsegs[] into softc. However, that would waste memory and make Tx/Rx speration hard when we want to sperate Tx/Rx handlers to optimize locking. - Fix dma map tracking used in Tx path. Previously it used the dma map of the last mbuf chain in ti_txeof() which was incorrect as ti(4) used dma map of the first mbuf chain when it loads a mbuf chain with bus_dmamap_load_mbuf(9). Correct the bug by introducing queues that keep track of active/inactive dma maps/mbuf chain. - Use ti_txcnt to check whether driver need to set watchdog timer instead of blidnly clearing the timer in ti_txeof(). - Remove the 3rd arg. of ti_encap(). Since ti(4) now caches the last descriptor index(ti_tx_saved_prodidx) used in Tx there is no need to pass it as a fuction arg. - Change data type of producer/consumer index to int from u_int16_t in order to remove implicit type conversions in Tx/Rx handlers. - Check interface queue before getting a mbuf chain to reduce locking overhead. - Check number of available Tx descriptores to be 16 or higher in ti_start(). This wouldn't protect Tx descriptor shortage but it would reduce number of bus_dmamap_unload(9) calls in ti_encap() when we are about to running out of Tx descriptors. - Command NIC to send packets ony when the driver really has packets enqueued. Previously it always set TI_MB_SENDPROD_IDX which would command NIC to DMA Tx descriptors into NIC local memory regardless of Tx descriptor changes. Reviewed by: scottl
2006-01-03 06:14:07 +00:00
struct ti_txdesc ti_txdesc[TI_TX_RING_CNT];
struct ti_txdq ti_txfreeq;
struct ti_txdq ti_txbusyq;
Overhaul bus_dma(9) usage in driver: - Don't use a single big DMA block for all rings. Create separate DMA area for each ring instead. Currently the following DMA areas are created: Event ring, standard RX ring, jumbo RX ring, RX return ring, hardware MAC statistics and producer/consumer status area. For Tigon II, mini RX ring and TX ring are additionally created. - Added missing bus_dmamap_sync(9) in various TX/RX paths. - TX ring is no longer created for Tigon 1 such that it saves more resources on Tigon 1. - Data sheet is not clear about alignment requirement of each ring so use 32 bytes alignment for normal DMA area but use 64 bytes alignment for jumbo RX ring where the extended RX descriptor size is 64 bytes. - For each TX/RX buffers use separate DMA tag(e.g. the size of a DMA segment, total size of DMA segments etc). - Tigon allows separate DMA area for event producer, RX return producer and TX consumer which is really cool feature. This means TX and RX path could be independently run in parallel. However ti(4) uses a single driver lock so it's meaningless to have separate DMA area for these producer/consumer such that this change creates a single status DMA area. - It seems Tigon has no limits on DMA address space and I also don't see any problem with that but old comments in driver indicates there could be issues on descriptors being located in 64bit region. Introduce a tunable, dev.ti.%d.dac, to disable using 64bit DMA in driver. The default is 0 which means it would use full 64bit DMA. If there are DMA issues, users can disable it by setting the tunable to 0. - Do not increase watchdog timer in ti_txeof(). Previously driver increased the watchdog timer whenever there are queued TX frames. - When stat ticks is set to 0, skip processing ti_stats_update(), avoiding bus_dmamap_sync(9) and updating if_collisions counter. - MTU does not include FCS bytes, replace it with ETHER_VLAN_ENCAP_LEN. With these changes, ti(4) should work on PAE environments. Many thanks to Jay Borkenhagen for remote hardware access.
2011-11-14 20:38:14 +00:00
bus_dma_tag_t ti_rx_return_ring_tag;
bus_dmamap_t ti_rx_return_ring_map;
bus_dma_tag_t ti_rx_std_ring_tag;
bus_dmamap_t ti_rx_std_ring_map;
bus_dma_tag_t ti_rx_std_tag;
struct mbuf *ti_rx_std_chain[TI_STD_RX_RING_CNT];
bus_dmamap_t ti_rx_std_maps[TI_STD_RX_RING_CNT];
bus_dmamap_t ti_rx_std_sparemap;
Overhaul bus_dma(9) usage in driver: - Don't use a single big DMA block for all rings. Create separate DMA area for each ring instead. Currently the following DMA areas are created: Event ring, standard RX ring, jumbo RX ring, RX return ring, hardware MAC statistics and producer/consumer status area. For Tigon II, mini RX ring and TX ring are additionally created. - Added missing bus_dmamap_sync(9) in various TX/RX paths. - TX ring is no longer created for Tigon 1 such that it saves more resources on Tigon 1. - Data sheet is not clear about alignment requirement of each ring so use 32 bytes alignment for normal DMA area but use 64 bytes alignment for jumbo RX ring where the extended RX descriptor size is 64 bytes. - For each TX/RX buffers use separate DMA tag(e.g. the size of a DMA segment, total size of DMA segments etc). - Tigon allows separate DMA area for event producer, RX return producer and TX consumer which is really cool feature. This means TX and RX path could be independently run in parallel. However ti(4) uses a single driver lock so it's meaningless to have separate DMA area for these producer/consumer such that this change creates a single status DMA area. - It seems Tigon has no limits on DMA address space and I also don't see any problem with that but old comments in driver indicates there could be issues on descriptors being located in 64bit region. Introduce a tunable, dev.ti.%d.dac, to disable using 64bit DMA in driver. The default is 0 which means it would use full 64bit DMA. If there are DMA issues, users can disable it by setting the tunable to 0. - Do not increase watchdog timer in ti_txeof(). Previously driver increased the watchdog timer whenever there are queued TX frames. - When stat ticks is set to 0, skip processing ti_stats_update(), avoiding bus_dmamap_sync(9) and updating if_collisions counter. - MTU does not include FCS bytes, replace it with ETHER_VLAN_ENCAP_LEN. With these changes, ti(4) should work on PAE environments. Many thanks to Jay Borkenhagen for remote hardware access.
2011-11-14 20:38:14 +00:00
bus_dma_tag_t ti_rx_jumbo_ring_tag;
bus_dmamap_t ti_rx_jumbo_ring_map;
bus_dma_tag_t ti_rx_jumbo_tag;
struct mbuf *ti_rx_jumbo_chain[TI_JUMBO_RX_RING_CNT];
bus_dmamap_t ti_rx_jumbo_maps[TI_JUMBO_RX_RING_CNT];
bus_dmamap_t ti_rx_jumbo_sparemap;
Overhaul bus_dma(9) usage in driver: - Don't use a single big DMA block for all rings. Create separate DMA area for each ring instead. Currently the following DMA areas are created: Event ring, standard RX ring, jumbo RX ring, RX return ring, hardware MAC statistics and producer/consumer status area. For Tigon II, mini RX ring and TX ring are additionally created. - Added missing bus_dmamap_sync(9) in various TX/RX paths. - TX ring is no longer created for Tigon 1 such that it saves more resources on Tigon 1. - Data sheet is not clear about alignment requirement of each ring so use 32 bytes alignment for normal DMA area but use 64 bytes alignment for jumbo RX ring where the extended RX descriptor size is 64 bytes. - For each TX/RX buffers use separate DMA tag(e.g. the size of a DMA segment, total size of DMA segments etc). - Tigon allows separate DMA area for event producer, RX return producer and TX consumer which is really cool feature. This means TX and RX path could be independently run in parallel. However ti(4) uses a single driver lock so it's meaningless to have separate DMA area for these producer/consumer such that this change creates a single status DMA area. - It seems Tigon has no limits on DMA address space and I also don't see any problem with that but old comments in driver indicates there could be issues on descriptors being located in 64bit region. Introduce a tunable, dev.ti.%d.dac, to disable using 64bit DMA in driver. The default is 0 which means it would use full 64bit DMA. If there are DMA issues, users can disable it by setting the tunable to 0. - Do not increase watchdog timer in ti_txeof(). Previously driver increased the watchdog timer whenever there are queued TX frames. - When stat ticks is set to 0, skip processing ti_stats_update(), avoiding bus_dmamap_sync(9) and updating if_collisions counter. - MTU does not include FCS bytes, replace it with ETHER_VLAN_ENCAP_LEN. With these changes, ti(4) should work on PAE environments. Many thanks to Jay Borkenhagen for remote hardware access.
2011-11-14 20:38:14 +00:00
bus_dma_tag_t ti_rx_mini_ring_tag;
bus_dmamap_t ti_rx_mini_ring_map;
bus_dma_tag_t ti_rx_mini_tag;
struct mbuf *ti_rx_mini_chain[TI_MINI_RX_RING_CNT];
bus_dmamap_t ti_rx_mini_maps[TI_MINI_RX_RING_CNT];
bus_dmamap_t ti_rx_mini_sparemap;
};
struct ti_type {
2011-11-04 19:12:07 +00:00
uint16_t ti_vid;
uint16_t ti_did;
const char *ti_name;
};
#define TI_HWREV_TIGON 0x01
#define TI_HWREV_TIGON_II 0x02
#define TI_TIMEOUT 1000
#define TI_TXCONS_UNSET 0xFFFF /* impossible value */
struct ti_mc_entry {
struct ether_addr mc_addr;
SLIST_ENTRY(ti_mc_entry) mc_entries;
};
At long last, commit the zero copy sockets code. MAKEDEV: Add MAKEDEV glue for the ti(4) device nodes. ti.4: Update the ti(4) man page to include information on the TI_JUMBO_HDRSPLIT and TI_PRIVATE_JUMBOS kernel options, and also include information about the new character device interface and the associated ioctls. man9/Makefile: Add jumbo.9 and zero_copy.9 man pages and associated links. jumbo.9: New man page describing the jumbo buffer allocator interface and operation. zero_copy.9: New man page describing the general characteristics of the zero copy send and receive code, and what an application author should do to take advantage of the zero copy functionality. NOTES: Add entries for ZERO_COPY_SOCKETS, TI_PRIVATE_JUMBOS, TI_JUMBO_HDRSPLIT, MSIZE, and MCLSHIFT. conf/files: Add uipc_jumbo.c and uipc_cow.c. conf/options: Add the 5 options mentioned above. kern_subr.c: Receive side zero copy implementation. This takes "disposable" pages attached to an mbuf, gives them to a user process, and then recycles the user's page. This is only active when ZERO_COPY_SOCKETS is turned on and the kern.ipc.zero_copy.receive sysctl variable is set to 1. uipc_cow.c: Send side zero copy functions. Takes a page written by the user and maps it copy on write and assigns it kernel virtual address space. Removes copy on write mapping once the buffer has been freed by the network stack. uipc_jumbo.c: Jumbo disposable page allocator code. This allocates (optionally) disposable pages for network drivers that want to give the user the option of doing zero copy receive. uipc_socket.c: Add kern.ipc.zero_copy.{send,receive} sysctls that are enabled if ZERO_COPY_SOCKETS is turned on. Add zero copy send support to sosend() -- pages get mapped into the kernel instead of getting copied if they meet size and alignment restrictions. uipc_syscalls.c:Un-staticize some of the sf* functions so that they can be used elsewhere. (uipc_cow.c) if_media.c: In the SIOCGIFMEDIA ioctl in ifmedia_ioctl(), avoid calling malloc() with M_WAITOK. Return an error if the M_NOWAIT malloc fails. The ti(4) driver and the wi(4) driver, at least, call this with a mutex held. This causes witness warnings for 'ifconfig -a' with a wi(4) or ti(4) board in the system. (I've only verified for ti(4)). ip_output.c: Fragment large datagrams so that each segment contains a multiple of PAGE_SIZE amount of data plus headers. This allows the receiver to potentially do page flipping on receives. if_ti.c: Add zero copy receive support to the ti(4) driver. If TI_PRIVATE_JUMBOS is not defined, it now uses the jumbo(9) buffer allocator for jumbo receive buffers. Add a new character device interface for the ti(4) driver for the new debugging interface. This allows (a patched version of) gdb to talk to the Tigon board and debug the firmware. There are also a few additional debugging ioctls available through this interface. Add header splitting support to the ti(4) driver. Tweak some of the default interrupt coalescing parameters to more useful defaults. Add hooks for supporting transmit flow control, but leave it turned off with a comment describing why it is turned off. if_tireg.h: Change the firmware rev to 12.4.11, since we're really at 12.4.11 plus fixes from 12.4.13. Add defines needed for debugging. Remove the ti_stats structure, it is now defined in sys/tiio.h. ti_fw.h: 12.4.11 firmware. ti_fw2.h: 12.4.11 firmware, plus selected fixes from 12.4.13, and my header splitting patches. Revision 12.4.13 doesn't handle 10/100 negotiation properly. (This firmware is the same as what was in the tree previously, with the addition of header splitting support.) sys/jumbo.h: Jumbo buffer allocator interface. sys/mbuf.h: Add a new external mbuf type, EXT_DISPOSABLE, to indicate that the payload buffer can be thrown away / flipped to a userland process. socketvar.h: Add prototype for socow_setup. tiio.h: ioctl interface to the character portion of the ti(4) driver, plus associated structure/type definitions. uio.h: Change prototype for uiomoveco() so that we'll know whether the source page is disposable. ufs_readwrite.c:Update for new prototype of uiomoveco(). vm_fault.c: In vm_fault(), check to see whether we need to do a page based copy on write fault. vm_object.c: Add a new function, vm_object_allocate_wait(). This does the same thing that vm_object allocate does, except that it gives the caller the opportunity to specify whether it should wait on the uma_zalloc() of the object structre. This allows vm objects to be allocated while holding a mutex. (Without generating WITNESS warnings.) vm_object_allocate() is implemented as a call to vm_object_allocate_wait() with the malloc flag set to M_WAITOK. vm_object.h: Add prototype for vm_object_allocate_wait(). vm_page.c: Add page-based copy on write setup, clear and fault routines. vm_page.h: Add page based COW function prototypes and variable in the vm_page structure. Many thanks to Drew Gallatin, who wrote the zero copy send and receive code, and to all the other folks who have tested and reviewed this code over the years.
2002-06-26 03:37:47 +00:00
typedef enum {
TI_FLAG_NONE = 0x00,
TI_FLAG_DEBUGING = 0x01,
TI_FLAG_WAIT_FOR_LINK = 0x02
} ti_flag_vals;
struct ti_softc {
device_t ti_dev;
struct ifnet *ti_ifp;
bus_space_handle_t ti_bhandle;
bus_space_tag_t ti_btag;
void *ti_intrhand;
struct resource *ti_irq;
struct resource *ti_res;
struct ifmedia ifmedia; /* media info */
2011-11-04 19:12:07 +00:00
uint8_t ti_hwrev; /* Tigon rev (1 or 2) */
uint8_t ti_copper; /* 1000baseTX card */
uint8_t ti_linkstat; /* Link state */
At long last, commit the zero copy sockets code. MAKEDEV: Add MAKEDEV glue for the ti(4) device nodes. ti.4: Update the ti(4) man page to include information on the TI_JUMBO_HDRSPLIT and TI_PRIVATE_JUMBOS kernel options, and also include information about the new character device interface and the associated ioctls. man9/Makefile: Add jumbo.9 and zero_copy.9 man pages and associated links. jumbo.9: New man page describing the jumbo buffer allocator interface and operation. zero_copy.9: New man page describing the general characteristics of the zero copy send and receive code, and what an application author should do to take advantage of the zero copy functionality. NOTES: Add entries for ZERO_COPY_SOCKETS, TI_PRIVATE_JUMBOS, TI_JUMBO_HDRSPLIT, MSIZE, and MCLSHIFT. conf/files: Add uipc_jumbo.c and uipc_cow.c. conf/options: Add the 5 options mentioned above. kern_subr.c: Receive side zero copy implementation. This takes "disposable" pages attached to an mbuf, gives them to a user process, and then recycles the user's page. This is only active when ZERO_COPY_SOCKETS is turned on and the kern.ipc.zero_copy.receive sysctl variable is set to 1. uipc_cow.c: Send side zero copy functions. Takes a page written by the user and maps it copy on write and assigns it kernel virtual address space. Removes copy on write mapping once the buffer has been freed by the network stack. uipc_jumbo.c: Jumbo disposable page allocator code. This allocates (optionally) disposable pages for network drivers that want to give the user the option of doing zero copy receive. uipc_socket.c: Add kern.ipc.zero_copy.{send,receive} sysctls that are enabled if ZERO_COPY_SOCKETS is turned on. Add zero copy send support to sosend() -- pages get mapped into the kernel instead of getting copied if they meet size and alignment restrictions. uipc_syscalls.c:Un-staticize some of the sf* functions so that they can be used elsewhere. (uipc_cow.c) if_media.c: In the SIOCGIFMEDIA ioctl in ifmedia_ioctl(), avoid calling malloc() with M_WAITOK. Return an error if the M_NOWAIT malloc fails. The ti(4) driver and the wi(4) driver, at least, call this with a mutex held. This causes witness warnings for 'ifconfig -a' with a wi(4) or ti(4) board in the system. (I've only verified for ti(4)). ip_output.c: Fragment large datagrams so that each segment contains a multiple of PAGE_SIZE amount of data plus headers. This allows the receiver to potentially do page flipping on receives. if_ti.c: Add zero copy receive support to the ti(4) driver. If TI_PRIVATE_JUMBOS is not defined, it now uses the jumbo(9) buffer allocator for jumbo receive buffers. Add a new character device interface for the ti(4) driver for the new debugging interface. This allows (a patched version of) gdb to talk to the Tigon board and debug the firmware. There are also a few additional debugging ioctls available through this interface. Add header splitting support to the ti(4) driver. Tweak some of the default interrupt coalescing parameters to more useful defaults. Add hooks for supporting transmit flow control, but leave it turned off with a comment describing why it is turned off. if_tireg.h: Change the firmware rev to 12.4.11, since we're really at 12.4.11 plus fixes from 12.4.13. Add defines needed for debugging. Remove the ti_stats structure, it is now defined in sys/tiio.h. ti_fw.h: 12.4.11 firmware. ti_fw2.h: 12.4.11 firmware, plus selected fixes from 12.4.13, and my header splitting patches. Revision 12.4.13 doesn't handle 10/100 negotiation properly. (This firmware is the same as what was in the tree previously, with the addition of header splitting support.) sys/jumbo.h: Jumbo buffer allocator interface. sys/mbuf.h: Add a new external mbuf type, EXT_DISPOSABLE, to indicate that the payload buffer can be thrown away / flipped to a userland process. socketvar.h: Add prototype for socow_setup. tiio.h: ioctl interface to the character portion of the ti(4) driver, plus associated structure/type definitions. uio.h: Change prototype for uiomoveco() so that we'll know whether the source page is disposable. ufs_readwrite.c:Update for new prototype of uiomoveco(). vm_fault.c: In vm_fault(), check to see whether we need to do a page based copy on write fault. vm_object.c: Add a new function, vm_object_allocate_wait(). This does the same thing that vm_object allocate does, except that it gives the caller the opportunity to specify whether it should wait on the uma_zalloc() of the object structre. This allows vm objects to be allocated while holding a mutex. (Without generating WITNESS warnings.) vm_object_allocate() is implemented as a call to vm_object_allocate_wait() with the malloc flag set to M_WAITOK. vm_object.h: Add prototype for vm_object_allocate_wait(). vm_page.c: Add page-based copy on write setup, clear and fault routines. vm_page.h: Add page based COW function prototypes and variable in the vm_page structure. Many thanks to Drew Gallatin, who wrote the zero copy send and receive code, and to all the other folks who have tested and reviewed this code over the years.
2002-06-26 03:37:47 +00:00
int ti_hdrsplit; /* enable header splitting */
Overhaul bus_dma(9) usage in driver: - Don't use a single big DMA block for all rings. Create separate DMA area for each ring instead. Currently the following DMA areas are created: Event ring, standard RX ring, jumbo RX ring, RX return ring, hardware MAC statistics and producer/consumer status area. For Tigon II, mini RX ring and TX ring are additionally created. - Added missing bus_dmamap_sync(9) in various TX/RX paths. - TX ring is no longer created for Tigon 1 such that it saves more resources on Tigon 1. - Data sheet is not clear about alignment requirement of each ring so use 32 bytes alignment for normal DMA area but use 64 bytes alignment for jumbo RX ring where the extended RX descriptor size is 64 bytes. - For each TX/RX buffers use separate DMA tag(e.g. the size of a DMA segment, total size of DMA segments etc). - Tigon allows separate DMA area for event producer, RX return producer and TX consumer which is really cool feature. This means TX and RX path could be independently run in parallel. However ti(4) uses a single driver lock so it's meaningless to have separate DMA area for these producer/consumer such that this change creates a single status DMA area. - It seems Tigon has no limits on DMA address space and I also don't see any problem with that but old comments in driver indicates there could be issues on descriptors being located in 64bit region. Introduce a tunable, dev.ti.%d.dac, to disable using 64bit DMA in driver. The default is 0 which means it would use full 64bit DMA. If there are DMA issues, users can disable it by setting the tunable to 0. - Do not increase watchdog timer in ti_txeof(). Previously driver increased the watchdog timer whenever there are queued TX frames. - When stat ticks is set to 0, skip processing ti_stats_update(), avoiding bus_dmamap_sync(9) and updating if_collisions counter. - MTU does not include FCS bytes, replace it with ETHER_VLAN_ENCAP_LEN. With these changes, ti(4) should work on PAE environments. Many thanks to Jay Borkenhagen for remote hardware access.
2011-11-14 20:38:14 +00:00
int ti_dac;
struct ti_ring_data ti_rdata; /* rings */
struct ti_chain_data ti_cdata; /* mbufs */
Overhaul bus_dma(9) usage in driver: - Don't use a single big DMA block for all rings. Create separate DMA area for each ring instead. Currently the following DMA areas are created: Event ring, standard RX ring, jumbo RX ring, RX return ring, hardware MAC statistics and producer/consumer status area. For Tigon II, mini RX ring and TX ring are additionally created. - Added missing bus_dmamap_sync(9) in various TX/RX paths. - TX ring is no longer created for Tigon 1 such that it saves more resources on Tigon 1. - Data sheet is not clear about alignment requirement of each ring so use 32 bytes alignment for normal DMA area but use 64 bytes alignment for jumbo RX ring where the extended RX descriptor size is 64 bytes. - For each TX/RX buffers use separate DMA tag(e.g. the size of a DMA segment, total size of DMA segments etc). - Tigon allows separate DMA area for event producer, RX return producer and TX consumer which is really cool feature. This means TX and RX path could be independently run in parallel. However ti(4) uses a single driver lock so it's meaningless to have separate DMA area for these producer/consumer such that this change creates a single status DMA area. - It seems Tigon has no limits on DMA address space and I also don't see any problem with that but old comments in driver indicates there could be issues on descriptors being located in 64bit region. Introduce a tunable, dev.ti.%d.dac, to disable using 64bit DMA in driver. The default is 0 which means it would use full 64bit DMA. If there are DMA issues, users can disable it by setting the tunable to 0. - Do not increase watchdog timer in ti_txeof(). Previously driver increased the watchdog timer whenever there are queued TX frames. - When stat ticks is set to 0, skip processing ti_stats_update(), avoiding bus_dmamap_sync(9) and updating if_collisions counter. - MTU does not include FCS bytes, replace it with ETHER_VLAN_ENCAP_LEN. With these changes, ti(4) should work on PAE environments. Many thanks to Jay Borkenhagen for remote hardware access.
2011-11-14 20:38:14 +00:00
#define ti_ev_prodidx ti_rdata.ti_status->ti_ev_prodidx_r
#define ti_return_prodidx ti_rdata.ti_status->ti_return_prodidx_r
#define ti_tx_considx ti_rdata.ti_status->ti_tx_considx_r
- Tx side bus_dmamap_load_mbuf_sg(9) support. This reduces bookkeeping requiried to keep consistent softc state before/after callback function invocation and supposed to be sligntly faster than previous one as it wouldn't incur callback overhead. With this change callback function was gone. - Decrease TI_MAXTXSEGS to 32 from 128. It seems that most mbuf chain length is less than 32 and it would be re-packed with m_defrag(9) if its chain length is larger than TI_MAXTXSEGS. This would protect ti(4) against possible kernel stack overflow when txsegs[] is put on stack. Alternatively, we can embed the txsegs[] into softc. However, that would waste memory and make Tx/Rx speration hard when we want to sperate Tx/Rx handlers to optimize locking. - Fix dma map tracking used in Tx path. Previously it used the dma map of the last mbuf chain in ti_txeof() which was incorrect as ti(4) used dma map of the first mbuf chain when it loads a mbuf chain with bus_dmamap_load_mbuf(9). Correct the bug by introducing queues that keep track of active/inactive dma maps/mbuf chain. - Use ti_txcnt to check whether driver need to set watchdog timer instead of blidnly clearing the timer in ti_txeof(). - Remove the 3rd arg. of ti_encap(). Since ti(4) now caches the last descriptor index(ti_tx_saved_prodidx) used in Tx there is no need to pass it as a fuction arg. - Change data type of producer/consumer index to int from u_int16_t in order to remove implicit type conversions in Tx/Rx handlers. - Check interface queue before getting a mbuf chain to reduce locking overhead. - Check number of available Tx descriptores to be 16 or higher in ti_start(). This wouldn't protect Tx descriptor shortage but it would reduce number of bus_dmamap_unload(9) calls in ti_encap() when we are about to running out of Tx descriptors. - Command NIC to send packets ony when the driver really has packets enqueued. Previously it always set TI_MB_SENDPROD_IDX which would command NIC to DMA Tx descriptors into NIC local memory regardless of Tx descriptor changes. Reviewed by: scottl
2006-01-03 06:14:07 +00:00
int ti_tx_saved_prodidx;
int ti_tx_saved_considx;
int ti_rx_saved_considx;
int ti_ev_saved_considx;
int ti_cmd_saved_prodidx;
int ti_std; /* current std ring head */
int ti_mini; /* current mini ring head */
int ti_jumbo; /* current jumo ring head */
SLIST_HEAD(__ti_mchead, ti_mc_entry) ti_mc_listhead;
2011-11-04 19:12:07 +00:00
uint32_t ti_stat_ticks;
uint32_t ti_rx_coal_ticks;
uint32_t ti_tx_coal_ticks;
uint32_t ti_rx_max_coal_bds;
uint32_t ti_tx_max_coal_bds;
uint32_t ti_tx_buf_ratio;
int ti_if_flags;
int ti_txcnt;
struct mtx ti_mtx;
struct callout ti_watchdog;
int ti_timer;
At long last, commit the zero copy sockets code. MAKEDEV: Add MAKEDEV glue for the ti(4) device nodes. ti.4: Update the ti(4) man page to include information on the TI_JUMBO_HDRSPLIT and TI_PRIVATE_JUMBOS kernel options, and also include information about the new character device interface and the associated ioctls. man9/Makefile: Add jumbo.9 and zero_copy.9 man pages and associated links. jumbo.9: New man page describing the jumbo buffer allocator interface and operation. zero_copy.9: New man page describing the general characteristics of the zero copy send and receive code, and what an application author should do to take advantage of the zero copy functionality. NOTES: Add entries for ZERO_COPY_SOCKETS, TI_PRIVATE_JUMBOS, TI_JUMBO_HDRSPLIT, MSIZE, and MCLSHIFT. conf/files: Add uipc_jumbo.c and uipc_cow.c. conf/options: Add the 5 options mentioned above. kern_subr.c: Receive side zero copy implementation. This takes "disposable" pages attached to an mbuf, gives them to a user process, and then recycles the user's page. This is only active when ZERO_COPY_SOCKETS is turned on and the kern.ipc.zero_copy.receive sysctl variable is set to 1. uipc_cow.c: Send side zero copy functions. Takes a page written by the user and maps it copy on write and assigns it kernel virtual address space. Removes copy on write mapping once the buffer has been freed by the network stack. uipc_jumbo.c: Jumbo disposable page allocator code. This allocates (optionally) disposable pages for network drivers that want to give the user the option of doing zero copy receive. uipc_socket.c: Add kern.ipc.zero_copy.{send,receive} sysctls that are enabled if ZERO_COPY_SOCKETS is turned on. Add zero copy send support to sosend() -- pages get mapped into the kernel instead of getting copied if they meet size and alignment restrictions. uipc_syscalls.c:Un-staticize some of the sf* functions so that they can be used elsewhere. (uipc_cow.c) if_media.c: In the SIOCGIFMEDIA ioctl in ifmedia_ioctl(), avoid calling malloc() with M_WAITOK. Return an error if the M_NOWAIT malloc fails. The ti(4) driver and the wi(4) driver, at least, call this with a mutex held. This causes witness warnings for 'ifconfig -a' with a wi(4) or ti(4) board in the system. (I've only verified for ti(4)). ip_output.c: Fragment large datagrams so that each segment contains a multiple of PAGE_SIZE amount of data plus headers. This allows the receiver to potentially do page flipping on receives. if_ti.c: Add zero copy receive support to the ti(4) driver. If TI_PRIVATE_JUMBOS is not defined, it now uses the jumbo(9) buffer allocator for jumbo receive buffers. Add a new character device interface for the ti(4) driver for the new debugging interface. This allows (a patched version of) gdb to talk to the Tigon board and debug the firmware. There are also a few additional debugging ioctls available through this interface. Add header splitting support to the ti(4) driver. Tweak some of the default interrupt coalescing parameters to more useful defaults. Add hooks for supporting transmit flow control, but leave it turned off with a comment describing why it is turned off. if_tireg.h: Change the firmware rev to 12.4.11, since we're really at 12.4.11 plus fixes from 12.4.13. Add defines needed for debugging. Remove the ti_stats structure, it is now defined in sys/tiio.h. ti_fw.h: 12.4.11 firmware. ti_fw2.h: 12.4.11 firmware, plus selected fixes from 12.4.13, and my header splitting patches. Revision 12.4.13 doesn't handle 10/100 negotiation properly. (This firmware is the same as what was in the tree previously, with the addition of header splitting support.) sys/jumbo.h: Jumbo buffer allocator interface. sys/mbuf.h: Add a new external mbuf type, EXT_DISPOSABLE, to indicate that the payload buffer can be thrown away / flipped to a userland process. socketvar.h: Add prototype for socow_setup. tiio.h: ioctl interface to the character portion of the ti(4) driver, plus associated structure/type definitions. uio.h: Change prototype for uiomoveco() so that we'll know whether the source page is disposable. ufs_readwrite.c:Update for new prototype of uiomoveco(). vm_fault.c: In vm_fault(), check to see whether we need to do a page based copy on write fault. vm_object.c: Add a new function, vm_object_allocate_wait(). This does the same thing that vm_object allocate does, except that it gives the caller the opportunity to specify whether it should wait on the uma_zalloc() of the object structre. This allows vm objects to be allocated while holding a mutex. (Without generating WITNESS warnings.) vm_object_allocate() is implemented as a call to vm_object_allocate_wait() with the malloc flag set to M_WAITOK. vm_object.h: Add prototype for vm_object_allocate_wait(). vm_page.c: Add page-based copy on write setup, clear and fault routines. vm_page.h: Add page based COW function prototypes and variable in the vm_page structure. Many thanks to Drew Gallatin, who wrote the zero copy send and receive code, and to all the other folks who have tested and reviewed this code over the years.
2002-06-26 03:37:47 +00:00
ti_flag_vals ti_flags;
uint8_t *ti_membuf;
uint8_t *ti_membuf2;
struct cdev *dev;
};
Change and clean the mutex lock interface. mtx_enter(lock, type) becomes: mtx_lock(lock) for sleep locks (MTX_DEF-initialized locks) mtx_lock_spin(lock) for spin locks (MTX_SPIN-initialized) similarily, for releasing a lock, we now have: mtx_unlock(lock) for MTX_DEF and mtx_unlock_spin(lock) for MTX_SPIN. We change the caller interface for the two different types of locks because the semantics are entirely different for each case, and this makes it explicitly clear and, at the same time, it rids us of the extra `type' argument. The enter->lock and exit->unlock change has been made with the idea that we're "locking data" and not "entering locked code" in mind. Further, remove all additional "flags" previously passed to the lock acquire/release routines with the exception of two: MTX_QUIET and MTX_NOSWITCH The functionality of these flags is preserved and they can be passed to the lock/unlock routines by calling the corresponding wrappers: mtx_{lock, unlock}_flags(lock, flag(s)) and mtx_{lock, unlock}_spin_flags(lock, flag(s)) for MTX_DEF and MTX_SPIN locks, respectively. Re-inline some lock acq/rel code; in the sleep lock case, we only inline the _obtain_lock()s in order to ensure that the inlined code fits into a cache line. In the spin lock case, we inline recursion and actually only perform a function call if we need to spin. This change has been made with the idea that we generally tend to avoid spin locks and that also the spin locks that we do have and are heavily used (i.e. sched_lock) do recurse, and therefore in an effort to reduce function call overhead for some architectures (such as alpha), we inline recursion for this case. Create a new malloc type for the witness code and retire from using the M_DEV type. The new type is called M_WITNESS and is only declared if WITNESS is enabled. Begin cleaning up some machdep/mutex.h code - specifically updated the "optimized" inlined code in alpha/mutex.h and wrote MTX_LOCK_SPIN and MTX_UNLOCK_SPIN asm macros for the i386/mutex.h as we presently need those. Finally, caught up to the interface changes in all sys code. Contributors: jake, jhb, jasone (in no particular order)
2001-02-09 06:11:45 +00:00
#define TI_LOCK(_sc) mtx_lock(&(_sc)->ti_mtx)
#define TI_UNLOCK(_sc) mtx_unlock(&(_sc)->ti_mtx)
#define TI_LOCK_ASSERT(_sc) mtx_assert(&(_sc)->ti_mtx, MA_OWNED)
/*
* Microchip Technology 24Cxx EEPROM control bytes
*/
#define EEPROM_CTL_READ 0xA1 /* 0101 0001 */
#define EEPROM_CTL_WRITE 0xA0 /* 0101 0000 */
/*
* Note that EEPROM_START leaves transmission enabled.
*/
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
#define EEPROM_START do { \
TI_SETBIT(sc, TI_MISC_LOCAL_CTL, TI_MLC_EE_CLK); /* Pull clock pin high */\
TI_SETBIT(sc, TI_MISC_LOCAL_CTL, TI_MLC_EE_DOUT); /* Set DATA bit to 1 */ \
TI_SETBIT(sc, TI_MISC_LOCAL_CTL, TI_MLC_EE_TXEN); /* Enable xmit to write bit */\
TI_CLRBIT(sc, TI_MISC_LOCAL_CTL, TI_MLC_EE_DOUT); /* Pull DATA bit to 0 again */\
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
TI_CLRBIT(sc, TI_MISC_LOCAL_CTL, TI_MLC_EE_CLK); /* Pull clock low again */ \
} while(0)
/*
* EEPROM_STOP ends access to the EEPROM and clears the ETXEN bit so
* that no further data can be written to the EEPROM I/O pin.
*/
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
#define EEPROM_STOP do { \
TI_CLRBIT(sc, TI_MISC_LOCAL_CTL, TI_MLC_EE_TXEN); /* Disable xmit */ \
TI_CLRBIT(sc, TI_MISC_LOCAL_CTL, TI_MLC_EE_DOUT); /* Pull DATA to 0 */ \
TI_SETBIT(sc, TI_MISC_LOCAL_CTL, TI_MLC_EE_CLK); /* Pull clock high */ \
TI_SETBIT(sc, TI_MISC_LOCAL_CTL, TI_MLC_EE_TXEN); /* Enable xmit */ \
TI_SETBIT(sc, TI_MISC_LOCAL_CTL, TI_MLC_EE_DOUT); /* Toggle DATA to 1 */ \
TI_CLRBIT(sc, TI_MISC_LOCAL_CTL, TI_MLC_EE_TXEN); /* Disable xmit. */ \
Bring big-endian architecture support for ti(4). . remove unnecessay header files after Scott's bus_dma(9) commit. . remove global variable tis which was introduced at the time of zero_copy(9) changes. The variable tis was not used at all. The same applyes to ti_links in softc so axe it. . deregister variables. . axe ti_vhandle and switch to use explicit register access for accessing NIC local memory. Creates three variants of ti_mem to read/write NIC local memory(ti_mem_read, ti_mem_write) and clearing NIC local memory(ti_mem_zero). This greatly enhances code readability and have ti(4) drop using shared memory scheme for Tigon 1. As Tigon 1 switched to use explicit register access for Tx, axe ti_tx_ring_nic/ti_cmd_ring in softc.(Tigon 2 used to host ring scheme which means there is no need to access NIC local memory via register access for Tx and NIC would DMA the modified Tx rings into its local memory.) [1] . introduce new macro TI_EVENT_*/TI_CMD_* to handle NIC envent/command. Instead of using bit fields assginment for accessing the event, use shift operations to set/get it. [1] . add additional check for valid DMA tags in ti_free_dmamaps(). . add missing bus_dmamap_sync/bus_dmamap_unload in ti_free_*_ring_*. . fix locking nits(MTX_RECURSE mutex) and make ti(4) MPSAFE. . change data type of ti_rdata_phys to bus_addr_t and don't blindly cast to uint32_t. . rearrange detach path and make ti(4) survive during device detach. . for Tigon 1, use explicit register access for checking Tx descriptors in ti_encap()/ti_txeof(). [1] . properly call bus_dmamap_sync(9) for updating statistics. . remove extra semicolon in ti_encap() . rewrite loading MAC address to work on strict-alignment architectures. . move TI_RD_OFF macro to if_tireg.h . axe ETHER_ALIGN as it's already defined in <net/ethernet.h>. . make macros immuine from expansion by adding parenthesis and do-while. . remove alpha specific hack as vtophys(9) is no longer used in ti(4) after Scott's bus_dma(9) fix. Reviewed by: scottl Obtained from: OpenBSD [1]
2005-12-28 02:57:19 +00:00
TI_CLRBIT(sc, TI_MISC_LOCAL_CTL, TI_MLC_EE_CLK); /* Pull clock low again */ \
} while(0)