summaryrefslogtreecommitdiffstats
path: root/include/asm-mips
Commit message (Collapse)AuthorAgeFilesLines
* [PATCH] Centralise definitions of sector_t and blkcnt_tMatthew Wilcox2006-12-041-10/+0
| | | | | | | | | | | | | | | | CONFIG_LBD and CONFIG_LSF are spread into asm/types.h for no particularly good reason. Centralising the definition in linux/types.h means that arch maintainers don't need to bother adding it, as well as fixing the problem with x86-64 users being asked to make a decision that has absolutely no effect. The H8/300 porters seem particularly confused since I'm not aware of any microcontrollers that need to support 2TB filesystems. Signed-off-by: Matthew Wilcox <matthew@wil.cx> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
* [MIPS] Cleanup memory barriers for weakly ordered systems.Ralf Baechle2006-12-046-209/+218
| | | | | | | Also the R4000 / R4600 LL/SC instructions imply a sync so no explicit sync needed. Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [MIPS] Fix atomic.h build errors.Ralf Baechle2006-12-041-1/+1
| | | | | | | | | For the definition of atomic64_t atomic.h was relying on <asm/types.h> having been included previously. Before changeset d89d8e0637a5e4e0a12e90c4bc934d0d4c335239 this was happening as a side effect of including <linux/spinlock.h>. Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [MIPS] Use SYSVIPC_COMPAT to fix various problems on N32Atsushi Nemoto2006-12-041-0/+68
| | | | | | | | | | | | | | N32 SysV IPC system calls should use 32-bit compatible code. arch/mips/kernel/linux32.c have similar compatible code for O32, but ipc/compat.c seems more complete. We can use it for both N32 and O32. This patch should fix these problems (and other possible problems): http://www.linux-mips.org/cgi-bin/mesg.cgi?a=linux-mips&i=1149188824.6986.6.camel%40diimka-laptop http://www.linux-mips.org/cgi-bin/mesg.cgi?a=linux-mips&i=44C6B829.8050508%40caviumnetworks.com Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [MIPS] klconfig add missing bracketMariusz Kozlowski2006-12-041-1/+1
| | | | | Signed-off-by: Mariusz Kozlowski <m.kozlowski@tuxland.pl> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [NET]: MIPS checksum annotations and cleanups.Al Viro2006-12-021-29/+26
| | | | | | | | | * sanitize prototypes, annotate * kill shift-by-16 in checksum calculations * htons->shift in l-e checksum calculations Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: David S. Miller <davem@davemloft.net>
* Merge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linusLinus Torvalds2006-12-0119-182/+170
|\ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus: (31 commits) [MIPS] Remove duplicate ISA DMA code for 0 DMA channel case. [MIPS] Remove unused definition of cpu_to_lelongp() [MIPS] Remove userspace proofing from <asm/bitops.h>. [MIPS] Remove old junk left from old atomic_lock. [MIPS] Use conditional traps for BUG_ON on MIPS II and better. [MIPS] mips HPT cleanup: make clocksource_mips public [MIPS] do_IRQ cleanup [MIPS] Avoid dupliate D-cache flush on R400C / R4400 SC and MC variants. [MIPS] Remove redundant r4k_blast_icache() calls [MIPS] Work around bogus gcc warnings. [MIPS] Fix double inclusions [MIPS] use generic_handle_irq, handle_level_irq, handle_percpu_irq [MIPS] IRQ cleanups [MIPS] mips hpt cleanup: get rid of mips_hpt_init [MIPS] PB1200: Remove duplicate definitions [MIPS] Fix alignment hole in struct cache_desc; shrink struct. [MIPS] Oprofile: kernel support for the R10000. [MIPS] Remove unused R10000 performance counter definitions. [MIPS] Add support for kexec [MIPS] Don't print presence of WAIT instruction on bootup. ...
| * [MIPS] Remove duplicate ISA DMA code for 0 DMA channel case.Ralf Baechle2006-11-301-0/+2
| | | | | | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] Remove unused definition of cpu_to_lelongp()Ralf Baechle2006-11-301-2/+0
| | | | | | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] Remove userspace proofing from <asm/bitops.h>.Ralf Baechle2006-11-301-42/+21
| | | | | | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] Remove old junk left from old atomic_lock.Ralf Baechle2006-11-301-9/+1
| | | | | | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] Use conditional traps for BUG_ON on MIPS II and better.Ralf Baechle2006-11-302-2/+12
| | | | | | | | | | | | | | This shaves of around 4kB and a few cycles for the average kernel that has CONFIG_BUG enabled. Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] mips HPT cleanup: make clocksource_mips publicAtsushi Nemoto2006-11-301-4/+4
| | | | | | | | | | | | | | | | Make clocksource_mips public and get rid of mips_hpt_read, mips_hpt_mask. Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] do_IRQ cleanupAtsushi Nemoto2006-11-301-6/+0
| | | | | | | | | | | | | | | | | | Now we have both function and macro version of do_IRQ() and the former is used only by DEC and non-preemptive kernel. This patch makes everyone use the macro version and removes the function version. Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] use generic_handle_irq, handle_level_irq, handle_percpu_irqAtsushi Nemoto2006-11-301-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Further incorporation of generic irq framework. Replacing __do_IRQ() by proper flow handler would make the irq handling path a bit simpler and faster. * use generic_handle_irq() instead of __do_IRQ(). * use handle_level_irq for obvious level-type irq chips. * use handle_percpu_irq for irqs marked as IRQ_PER_CPU. * setup .eoi routine for irq chips possibly used with handle_percpu_irq. Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] IRQ cleanupsAtsushi Nemoto2006-11-301-2/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This is a big irq cleanup patch. * Use set_irq_chip() to register irq_chip. * Initialize .mask, .unmask, .mask_ack field. Functions for these method are already exist in most case. * Do not initialize .startup, .shutdown, .enable, .disable fields if default routines provided by irq_chip_set_defaults() were suitable. * Remove redundant irq_desc initializations. * Remove unnecessary local_irq_save/local_irq_restore, spin_lock. With this cleanup, it would be easy to switch to slightly lightwait irq flow handlers (handle_level_irq(), etc.) instead of __do_IRQ(). Though whole this patch is quite large, changes in each irq_chip are not quite simple. Please review and test on your platform. Thanks. Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] mips hpt cleanup: get rid of mips_hpt_initAtsushi Nemoto2006-11-301-1/+0
| | | | | | | | | | | | | | | | | | Currently nobody outside time.c require mips_hpt_init(). Remove it and call c0_hpt_timer_init() directly if R4k counter was used for timer interrupt. Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] Fix alignment hole in struct cache_desc; shrink struct.Ralf Baechle2006-11-301-5/+5
| | | | | | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] Remove unused R10000 performance counter definitions.Ralf Baechle2006-11-301-56/+0
| | | | | | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] Add support for kexecNicolas Schichan2006-11-301-0/+32
| | | | | | | | | | | | | | | | | | A tiny userland application loading the kernel and invoking kexec_load for mips is available here: http://chac.le-poulpe.net/~nico/kexec/kexec-2006-10-18.tar.gz Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] Load modules to CKSEG0 if CONFIG_BUILD_ELF64=nAtsushi Nemoto2006-11-301-0/+13
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This is a patch to load 64-bit modules to CKSEG0 so that can be compiled with -msym32 option. This makes each module ~10% smaller. * introduce MODULE_START and MODULE_END * custom module_alloc() * PGD for modules * change XTLB refill handler synthesizer * enable -msym32 for modules again (revert ca78b1a5c6a6e70e052d3ea253828e49b5d07c8a) New XTLB refill handler looks like this: 80000080 dmfc0 k0,C0_BADVADDR 80000084 bltz k0,800000e4 # goto l_module_alloc 80000088 lui k1,0x8046 # %high(pgd_current) 8000008c ld k1,24600(k1) # %low(pgd_current) 80000090 dsrl k0,k0,0x1b # l_vmalloc_done: 80000094 andi k0,k0,0x1ff8 80000098 daddu k1,k1,k0 8000009c dmfc0 k0,C0_BADVADDR 800000a0 ld k1,0(k1) 800000a4 dsrl k0,k0,0x12 800000a8 andi k0,k0,0xff8 800000ac daddu k1,k1,k0 800000b0 dmfc0 k0,C0_XCONTEXT 800000b4 ld k1,0(k1) 800000b8 andi k0,k0,0xff0 800000bc daddu k1,k1,k0 800000c0 ld k0,0(k1) 800000c4 ld k1,8(k1) 800000c8 dsrl k0,k0,0x6 800000cc mtc0 k0,C0_ENTRYLO0 800000d0 dsrl k1,k1,0x6 800000d4 mtc0 k1,C0_ENTRYL01 800000d8 nop 800000dc tlbwr 800000e0 eret 800000e4 dsll k1,k0,0x2 # l_module_alloc: 800000e8 bgez k1,80000008 # goto l_vmalloc 800000ec lui k1,0xc000 800000f0 dsubu k0,k0,k1 800000f4 lui k1,0x8046 # %high(module_pg_dir) 800000f8 beq zero,zero,80000000 800000fc nop 80000000 beq zero,zero,80000090 # goto l_vmalloc_done 80000004 daddiu k1,k1,0x4000 80000008 dsll32 k1,k1,0x0 # l_vmalloc: 8000000c dsubu k0,k0,k1 80000010 beq zero,zero,80000090 # goto l_vmalloc_done 80000014 lui k1,0x8046 # %high(swapper_pg_dir) Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] Rewrite GALILEO_INL/GALILEO_OUTL to GT_READ/GT_WRITEYoichi Yuasa2006-11-303-28/+43
| | | | | | | | | | | | | | | | This patch has rewritten GALILEO_INL/GALILEO_OUTL using GT_READ/GT_WRITE. This patch tested on Cobalt Qube2. Signed-off-by: Yoichi Yuasa <yoichi_yuasa@tripeaks.co.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] Use "long" for 64-bit values on 64-bit kernel.Atsushi Nemoto2006-11-301-18/+22
| | | | | | | | | | | | | | This would get rid of some warnings about "long" vs. "long long". Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] Introduce __pa_symbol()Franck Bui-Huu2006-11-301-0/+1
| | | | | | | | | | | | | | | | | | | | This patch introduces __pa_symbol() macro which should be used to calculate the physical address of kernel symbols. It also relies on RELOC_HIDE() to avoid any compiler's oddities when doing arithmetics on symbols. Signed-off-by: Franck Bui-Huu <fbuihuu@gmail.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] Make __pa() aware of XKPHYS/CKSEG0 address mix for 64 bit kernelsFranck Bui-Huu2006-11-301-2/+7
| | | | | | | | | | | | | | | | | | | | During early boot mem init, some configs couldn't use __pa() to convert virtual into physical addresses. Specially for 64 bit kernel cases when CONFIG_BUILD_ELF64=n. This patch make __pa() work for _all_ configs and thus make CPHYSADDR() useless. Signed-off-by: Franck Bui-Huu <fbuihuu@gmail.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] page.h: remove __pa() usages.Franck Bui-Huu2006-11-303-4/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | __pa() was used by virt_to_page() and virt_addr_valid(). These latter are used when kernel is initialised so __pa() is not appropriate, we use virt_to_phys() instead. Futhermore __pa() is going to take care of CKSEG0/XKPHYS address mix for 64 bit kernels. This makes __pa() more complex than virt_to_phys() and this extra work is not needed by virt_to_page() and virt_addr_valid(). Eventually it consolidates virt_to_phys() prototype by making its argument 'const'. this avoids some warnings that was due to some virt_to_page() usages which pass const pointer. Signed-off-by: Franck Bui-Huu <fbuihuu@gmail.com> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* | Driver core: add dev_archdata to struct deviceBenjamin Herrenschmidt2006-12-011-0/+7
|/ | | | | | | | | | | | | | Add arch specific dev_archdata to struct device Adds an arch specific struct dev_arch to struct device. This enables architecture to add specific fields to every device in the system, like DMA operation pointers, NUMA node ID, firmware specific data, etc... Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Acked-by: Andi Kleen <ak@suse.de> Acked-By: David Howells <dhowells@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
* [PATCH] make au1xxx-ide compile againManuel Lauss2006-11-221-4/+2
| | | | | | | | | | | | | The Au1xx IDE controller driver doesn't compile: CC drivers/ide/mips/au1xxx-ide.o /linux-2.6.19-rc6-work/drivers/ide/mips/au1xxx-ide.c:480: error: conflicting types for 'auide_ddma_tx_callback' include2/asm/mach-au1x00/au1xxx_ide.h:174: error: previous declaration of 'auide_ddma_tx_callback' was here /linux-2.6.19-rc6-work/drivers/ide/mips/au1xxx-ide.c:486: error: conflicting types for 'auide_ddma_rx_callback' include2/asm/mach-au1x00/au1xxx_ide.h:176: error: previous declaration of 'auide_ddma_rx_callback' was here Signed-off-by: Manuel Lauss <mano@roarinelk.homelinux.net> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [MIPS] Make irq number allocator generally available for fixing EV64120.Ralf Baechle2006-11-061-0/+4
| | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [MIPS] Don't use R10000 llsc workaround version for all llsc-full processors.Ralf Baechle2006-11-021-1/+1
| | | | | | Found and original patch by bile@landofbile.com. Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [MIPS] 16K & 64K page size fixesRalf Baechle2006-11-013-2/+4
| | | | | | Derived from Peter Watkins <treestem@gmail.com>'s work. Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [MIPS] Fixup migration to GENERIC_TIMEAtsushi Nemoto2006-10-313-30/+3
| | | | | | | | | | | | | | | | | | | | | Since we already moved to GENERIC_TIME, we should implement alternatives of old do_gettimeoffset routines to get sub-jiffies resolution from gettimeofday(). This patch includes: * MIPS clocksource support (based on works by Manish Lachwani). * remove unused gettimeoffset routines and related codes. * remove unised 64bit do_div64_32(). * simplify mips_hpt_init. (no argument needed, __init tag) * simplify c0_hpt_timer_init. (no need to write to c0_count) * remove some hpt_init routines. * mips_hpt_mask variable to specify bitmask of hpt value. * convert jmr3927_do_gettimeoffset to jmr3927_hpt_read. * convert ip27_do_gettimeoffset to ip27_hpt_read. * convert bcm1480_do_gettimeoffset to bcm1480_hpt_read. * simplify sb1250 hpt functions. (no need to subtract and shift) Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [MIPS] VSMP: Synchronize cp0 counters on bootup.Ralf Baechle2006-10-311-0/+2
| | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [MIPS] MIPS doesn't need compat_sys_getdents.Ralf Baechle2006-10-301-0/+1
| | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [MIPS] Wire up getcpu(2) and epoll_wait(2) syscalls.Ralf Baechle2006-10-301-6/+12
| | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [MIPS] Fix aliasing bug in copy_to_user_page / copy_from_user_pageRalf Baechle2006-10-212-18/+15
| | | | | | | | | | | | | | | | The current implementation uses a sequence of a cacheflush and a copy. This is racy in case of a multithreaded debuggee and renders GDB virtually unusable. Aside this fixes a performance hog rendering access to /proc/cmdline very slow and resulting in a enough cache stalls for the 34K AP/SP programming model to make the bare metal code on the non-Linux VPE miss RT deadlines. The main part of this patch was originally written by Ralf Baechle; Atushi Nemoto did the the debugging. Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [MIPS] Fix iounmap argument to const volatile.Ralf Baechle2006-10-191-2/+2
| | | | | | | | | | | | With the existing prototype the following code: const void __iomem *io = ioremap(); x = readb(io); iounmap(io); did result in a warning. Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [MIPS] Reserve syscall numbers for kexec_load.Ralf Baechle2006-10-191-6/+9
| | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [MIPS] More vr41xx pt_regs fixupsYoichi Yuasa2006-10-191-1/+1
| | | | | Signed-off-by: Yoichi Yuasa <yoichi_yuasa@tripeaks.co.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* Merge branch 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linusLinus Torvalds2006-10-113-56/+22
|\ | | | | | | | | | | | | | | | | | | | | | | | | * 'upstream' of git://ftp.linux-mips.org/pub/scm/upstream-linus: [MIPS] Pass NULL not 0 for pointer value. [MIPS] IP27: Make declaration of setup_replication_mask a proper prototype. [MIPS] BigSur: More useful defconfig. [MIPS] Cleanup definitions of speed_t and tcflag_t. [MIPS] Fix compilation warnings in arch/mips/sibyte/bcm1480/smp.c [MIPS] Optimize and cleanup get_saved_sp, set_saved_sp [MIPS] <asm/irq.h> does not need pt_regs anymore. [MIPS] Workaround for bug in gcc -EB / -EL options. [MIPS] Fix timer setup for Jazz
| * [MIPS] Cleanup definitions of speed_t and tcflag_t.Ralf Baechle2006-10-111-9/+3
| | | | | | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] Optimize and cleanup get_saved_sp, set_saved_spAtsushi Nemoto2006-10-111-45/+19
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | If CONFIG_BUILD_ELF64 was not selected and gcc had -msym32 option (i.e. 4.0 or newer), there is no point to use %highest, %higher for kernel symbols. This patch also fixes 64-bit SMTC version of get_saved_sp() which is broken but harmless since there is no such CPUs for now. A bonus is set_saved_sp() and SMP version of get_saved_sp() are more readable now. Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
| * [MIPS] <asm/irq.h> does not need pt_regs anymore.Atsushi Nemoto2006-10-111-2/+0
| | | | | | | | | | Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* | [PATCH] Consolidate check_signatureMatthew Wilcox2006-10-111-26/+0
|/ | | | | | | | | | | There's nothing arch-specific about check_signature(), so move it to <linux/io.h>. Use a cross between the Alpha and i386 implementations as the generic one. Signed-off-by: Matthew Wilcox <willy@parisc-linux.org> Acked-by: Alan Cox <alan@redhat.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
* [MIPS] Make sure cpu_has_fpu is used only in atomic contextAtsushi Nemoto2006-10-091-2/+4
| | | | | | | | Make sure cpu_has_fpu (which uses smp_processor_id()) is used only in atomic context. Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp> Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [MIPS] Alchemy: nuke usbdev; it's useless as is ...Ralf Baechle2006-10-091-73/+0
| | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* IRQ: Use the new typedef for interrupt handler function pointersDavid Howells2006-10-091-1/+1
| | | | | | | | | | | | | | | | Use the new typedef for interrupt handler function pointers rather than actually spelling out the full thing each time. This was scripted with the following small shell script: #!/bin/sh egrep -nHrl -e 'irqreturn_t[ ]*[(][*]' $* | while read i do echo $i perl -pi -e 's/irqreturn_t\s*[(]\s*[*]\s*([_a-zA-Z0-9]*)\s*[)]\s*[(]\s*int\s*,\s*void\s*[*]\s*[)]/irq_handler_t \1/g' $i || exit $? done Signed-Off-By: David Howells <dhowells@redhat.com>
* [MIPS] DEC: pt_regs fixes for buserror handlersRalf Baechle2006-10-093-6/+3
| | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* [MIPS] Complete fixes after removal of pt_regs argument to int handlers.Ralf Baechle2006-10-088-15/+31
| | | | Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
* IRQ: Maintain regs pointer globally rather than passing to IRQ handlersDavid Howells2006-10-052-2/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
OpenPOWER on IntegriCloud