Re: [RESEND 2/3] powerpc/memcpy: Add memcpy_mcsafe for pmem
On Wed, May 2, 2018 at 6:57 AM, Dan Williamswrote: > On Thu, Apr 5, 2018 at 8:00 AM, Dan Williams wrote: >> On Wed, Apr 4, 2018 at 11:45 PM, Nicholas Piggin wrote: > [,,] >>> What's the problem with just counting bytes copied like usercopy -- >>> why is that harder than cacheline accuracy? >>> I'd rather implement the existing interface and port/support the new interface as it becomes available >>> >>> Fair enough. >> >> I have patches already in progress to change the interface. My >> preference is to hold off on adding a new implementation that will >> need to be immediately reworked. When I say "immediate" I mean that >> should be able to post what I have for review within the next few >> days. >> >> Whether this is all too late for 4.17 is another question... > > Here is the x86 version of a 'bytes remaining' memcpy_mcsafe() implemenation: > > https://lists.01.org/pipermail/linux-nvdimm/2018-May/015548.html Thanks for the heads up! I'll work on the implementation for powerpc. Balbir Singh.
Re: [RESEND 2/3] powerpc/memcpy: Add memcpy_mcsafe for pmem
On Thu, Apr 5, 2018 at 8:00 AM, Dan Williamswrote: > On Wed, Apr 4, 2018 at 11:45 PM, Nicholas Piggin wrote: [,,] >> What's the problem with just counting bytes copied like usercopy -- >> why is that harder than cacheline accuracy? >> >>> I'd rather implement the existing interface and port/support the new >>> interface >>> as it becomes available >> >> Fair enough. > > I have patches already in progress to change the interface. My > preference is to hold off on adding a new implementation that will > need to be immediately reworked. When I say "immediate" I mean that > should be able to post what I have for review within the next few > days. > > Whether this is all too late for 4.17 is another question... Here is the x86 version of a 'bytes remaining' memcpy_mcsafe() implemenation: https://lists.01.org/pipermail/linux-nvdimm/2018-May/015548.html
RE: [RESEND 2/3] powerpc/memcpy: Add memcpy_mcsafe for pmem
> I thought the cache-aligned might make sense, since usually we'd expect the > failure to be at a cache-line level, but our copy_tofrom_user does accurate > accounting That's one of the wrinkles in the current x86 memcpy_mcsafe(). It starts by checking alignment of the source address, and moves a byte at a time until it is 8-byte aligned. We do this because current x86 implementations do not gracefully handle an unaligned read that spans from a good cache-line into a poisoned one. This is different from copy_tofrom_user which aligns the destination for speed reasons (unaligned reads have a lower penalty than unaligned writes). -Tony
Re: [RESEND 2/3] powerpc/memcpy: Add memcpy_mcsafe for pmem
On Fri, Apr 6, 2018 at 11:26 AM, Nicholas Pigginwrote: > On Thu, 05 Apr 2018 16:40:26 -0400 > Jeff Moyer wrote: > >> Nicholas Piggin writes: >> >> > On Thu, 5 Apr 2018 15:53:07 +1000 >> > Balbir Singh wrote: >> >> I'm thinking about it, I wonder what "bytes remaining" mean in pmem >> >> context >> >> in the context of a machine check exception. Also, do we want to be byte >> >> accurate or cache-line accurate for the bytes remaining? The former is >> >> much >> >> easier than the latter :) >> > >> > The ideal would be a linear measure of how much of your copy reached >> > (or can reach) non-volatile storage with nothing further copied. You >> > may have to allow for some relaxing of the semantics depending on >> > what the architecture can support. >> >> I think you've got that backwards. memcpy_mcsafe is used to copy *from* >> persistent memory. The idea is to catch errors when reading pmem, not >> writing to it. >> I know the comment in x86 says posted writes and cares for only loads, but I don't see why both sides should not be handled. >> > What's the problem with just counting bytes copied like usercopy -- >> > why is that harder than cacheline accuracy? >> >> He said the former (i.e. bytes) is easier. So, I think you're on the >> same page. :) > > Oh well that makes a lot more sense in my mind now, thanks :) I thought the cache-aligned might make sense, since usually we'd expect the failure to be at a cache-line level, but our copy_tofrom_user does accurate accounting Balbir Singh.
Re: [RESEND 2/3] powerpc/memcpy: Add memcpy_mcsafe for pmem
On Thu, 05 Apr 2018 16:40:26 -0400 Jeff Moyerwrote: > Nicholas Piggin writes: > > > On Thu, 5 Apr 2018 15:53:07 +1000 > > Balbir Singh wrote: > >> I'm thinking about it, I wonder what "bytes remaining" mean in pmem context > >> in the context of a machine check exception. Also, do we want to be byte > >> accurate or cache-line accurate for the bytes remaining? The former is much > >> easier than the latter :) > > > > The ideal would be a linear measure of how much of your copy reached > > (or can reach) non-volatile storage with nothing further copied. You > > may have to allow for some relaxing of the semantics depending on > > what the architecture can support. > > I think you've got that backwards. memcpy_mcsafe is used to copy *from* > persistent memory. The idea is to catch errors when reading pmem, not > writing to it. > > > What's the problem with just counting bytes copied like usercopy -- > > why is that harder than cacheline accuracy? > > He said the former (i.e. bytes) is easier. So, I think you're on the > same page. :) Oh well that makes a lot more sense in my mind now, thanks :)
Re: [RESEND 2/3] powerpc/memcpy: Add memcpy_mcsafe for pmem
Nicholas Pigginwrites: > On Thu, 5 Apr 2018 15:53:07 +1000 > Balbir Singh wrote: >> I'm thinking about it, I wonder what "bytes remaining" mean in pmem context >> in the context of a machine check exception. Also, do we want to be byte >> accurate or cache-line accurate for the bytes remaining? The former is much >> easier than the latter :) > > The ideal would be a linear measure of how much of your copy reached > (or can reach) non-volatile storage with nothing further copied. You > may have to allow for some relaxing of the semantics depending on > what the architecture can support. I think you've got that backwards. memcpy_mcsafe is used to copy *from* persistent memory. The idea is to catch errors when reading pmem, not writing to it. > What's the problem with just counting bytes copied like usercopy -- > why is that harder than cacheline accuracy? He said the former (i.e. bytes) is easier. So, I think you're on the same page. :) Cheers, Jeff
Re: [RESEND 2/3] powerpc/memcpy: Add memcpy_mcsafe for pmem
On Wed, Apr 4, 2018 at 11:45 PM, Nicholas Pigginwrote: > On Thu, 5 Apr 2018 15:53:07 +1000 > Balbir Singh wrote: > >> On Thu, 5 Apr 2018 15:04:05 +1000 >> Nicholas Piggin wrote: >> >> > On Wed, 4 Apr 2018 20:00:52 -0700 >> > Dan Williams wrote: >> > >> > > [ adding Matthew, Christoph, and Tony ] >> > > >> > > On Wed, Apr 4, 2018 at 4:57 PM, Nicholas Piggin >> > > wrote: >> > > > On Thu, 5 Apr 2018 09:19:42 +1000 >> > > > Balbir Singh wrote: >> > > > >> > > >> The pmem infrastructure uses memcpy_mcsafe in the pmem >> > > >> layer so as to convert machine check excpetions into >> > > >> a return value on failure in case a machine check >> > > >> exception is encoutered during the memcpy. >> > > >> >> > > >> This patch largely borrows from the copyuser_power7 >> > > >> logic and does not add the VMX optimizations, largely >> > > >> to keep the patch simple. If needed those optimizations >> > > >> can be folded in. >> > > > >> > > > So memcpy_mcsafe doesn't return number of bytes copied? >> > > > Huh, well that makes it simple. >> > > >> > > Well, not in current kernels, but we need to add that support or >> > > remove the direct call to copy_to_iter() in fs/dax.c. I'm looking >> > > right now to add "bytes remaining" support to the x86 memcpy_mcsafe(), >> > > but for copy_to_user we also need to handle bytes remaining for write >> > > faults. That fix is hopefully something that can land in an early >> > > 4.17-rc, but it won't be ready for -rc1. >> > >> > I wonder if the powerpc implementation should just go straight to >> > counting bytes. Backporting to this interface would be trivial, but >> > it would just mean there's only one variant of the code to support. >> > That's up to Balbir though. >> > >> >> I'm thinking about it, I wonder what "bytes remaining" mean in pmem context >> in the context of a machine check exception. Also, do we want to be byte >> accurate or cache-line accurate for the bytes remaining? The former is much >> easier than the latter :) > > The ideal would be a linear measure of how much of your copy reached > (or can reach) non-volatile storage with nothing further copied. You > may have to allow for some relaxing of the semantics depending on > what the architecture can support. > > What's the problem with just counting bytes copied like usercopy -- > why is that harder than cacheline accuracy? > >> I'd rather implement the existing interface and port/support the new >> interface >> as it becomes available > > Fair enough. I have patches already in progress to change the interface. My preference is to hold off on adding a new implementation that will need to be immediately reworked. When I say "immediate" I mean that should be able to post what I have for review within the next few days. Whether this is all too late for 4.17 is another question...
Re: [RESEND 2/3] powerpc/memcpy: Add memcpy_mcsafe for pmem
On Thu, 5 Apr 2018 15:53:07 +1000 Balbir Singhwrote: > On Thu, 5 Apr 2018 15:04:05 +1000 > Nicholas Piggin wrote: > > > On Wed, 4 Apr 2018 20:00:52 -0700 > > Dan Williams wrote: > > > > > [ adding Matthew, Christoph, and Tony ] > > > > > > On Wed, Apr 4, 2018 at 4:57 PM, Nicholas Piggin > > > wrote: > > > > On Thu, 5 Apr 2018 09:19:42 +1000 > > > > Balbir Singh wrote: > > > > > > > >> The pmem infrastructure uses memcpy_mcsafe in the pmem > > > >> layer so as to convert machine check excpetions into > > > >> a return value on failure in case a machine check > > > >> exception is encoutered during the memcpy. > > > >> > > > >> This patch largely borrows from the copyuser_power7 > > > >> logic and does not add the VMX optimizations, largely > > > >> to keep the patch simple. If needed those optimizations > > > >> can be folded in. > > > > > > > > So memcpy_mcsafe doesn't return number of bytes copied? > > > > Huh, well that makes it simple. > > > > > > Well, not in current kernels, but we need to add that support or > > > remove the direct call to copy_to_iter() in fs/dax.c. I'm looking > > > right now to add "bytes remaining" support to the x86 memcpy_mcsafe(), > > > but for copy_to_user we also need to handle bytes remaining for write > > > faults. That fix is hopefully something that can land in an early > > > 4.17-rc, but it won't be ready for -rc1. > > > > I wonder if the powerpc implementation should just go straight to > > counting bytes. Backporting to this interface would be trivial, but > > it would just mean there's only one variant of the code to support. > > That's up to Balbir though. > > > > I'm thinking about it, I wonder what "bytes remaining" mean in pmem context > in the context of a machine check exception. Also, do we want to be byte > accurate or cache-line accurate for the bytes remaining? The former is much > easier than the latter :) The ideal would be a linear measure of how much of your copy reached (or can reach) non-volatile storage with nothing further copied. You may have to allow for some relaxing of the semantics depending on what the architecture can support. What's the problem with just counting bytes copied like usercopy -- why is that harder than cacheline accuracy? > I'd rather implement the existing interface and port/support the new interface > as it becomes available Fair enough. Thanks, Nick
Re: [RESEND 2/3] powerpc/memcpy: Add memcpy_mcsafe for pmem
On Thu, 5 Apr 2018 15:04:05 +1000 Nicholas Pigginwrote: > On Wed, 4 Apr 2018 20:00:52 -0700 > Dan Williams wrote: > > > [ adding Matthew, Christoph, and Tony ] > > > > On Wed, Apr 4, 2018 at 4:57 PM, Nicholas Piggin wrote: > > > On Thu, 5 Apr 2018 09:19:42 +1000 > > > Balbir Singh wrote: > > > > > >> The pmem infrastructure uses memcpy_mcsafe in the pmem > > >> layer so as to convert machine check excpetions into > > >> a return value on failure in case a machine check > > >> exception is encoutered during the memcpy. > > >> > > >> This patch largely borrows from the copyuser_power7 > > >> logic and does not add the VMX optimizations, largely > > >> to keep the patch simple. If needed those optimizations > > >> can be folded in. > > > > > > So memcpy_mcsafe doesn't return number of bytes copied? > > > Huh, well that makes it simple. > > > > Well, not in current kernels, but we need to add that support or > > remove the direct call to copy_to_iter() in fs/dax.c. I'm looking > > right now to add "bytes remaining" support to the x86 memcpy_mcsafe(), > > but for copy_to_user we also need to handle bytes remaining for write > > faults. That fix is hopefully something that can land in an early > > 4.17-rc, but it won't be ready for -rc1. > > I wonder if the powerpc implementation should just go straight to > counting bytes. Backporting to this interface would be trivial, but > it would just mean there's only one variant of the code to support. > That's up to Balbir though. > I'm thinking about it, I wonder what "bytes remaining" mean in pmem context in the context of a machine check exception. Also, do we want to be byte accurate or cache-line accurate for the bytes remaining? The former is much easier than the latter :) I'd rather implement the existing interface and port/support the new interface as it becomes available Balbir Singh.
Re: [RESEND 2/3] powerpc/memcpy: Add memcpy_mcsafe for pmem
On Wed, 4 Apr 2018 20:00:52 -0700 Dan Williamswrote: > [ adding Matthew, Christoph, and Tony ] > > On Wed, Apr 4, 2018 at 4:57 PM, Nicholas Piggin wrote: > > On Thu, 5 Apr 2018 09:19:42 +1000 > > Balbir Singh wrote: > > > >> The pmem infrastructure uses memcpy_mcsafe in the pmem > >> layer so as to convert machine check excpetions into > >> a return value on failure in case a machine check > >> exception is encoutered during the memcpy. > >> > >> This patch largely borrows from the copyuser_power7 > >> logic and does not add the VMX optimizations, largely > >> to keep the patch simple. If needed those optimizations > >> can be folded in. > > > > So memcpy_mcsafe doesn't return number of bytes copied? > > Huh, well that makes it simple. > > Well, not in current kernels, but we need to add that support or > remove the direct call to copy_to_iter() in fs/dax.c. I'm looking > right now to add "bytes remaining" support to the x86 memcpy_mcsafe(), > but for copy_to_user we also need to handle bytes remaining for write > faults. That fix is hopefully something that can land in an early > 4.17-rc, but it won't be ready for -rc1. I wonder if the powerpc implementation should just go straight to counting bytes. Backporting to this interface would be trivial, but it would just mean there's only one variant of the code to support. That's up to Balbir though. Thanks, Nick
Re: [RESEND 2/3] powerpc/memcpy: Add memcpy_mcsafe for pmem
[ adding Matthew, Christoph, and Tony ] On Wed, Apr 4, 2018 at 4:57 PM, Nicholas Pigginwrote: > On Thu, 5 Apr 2018 09:19:42 +1000 > Balbir Singh wrote: > >> The pmem infrastructure uses memcpy_mcsafe in the pmem >> layer so as to convert machine check excpetions into >> a return value on failure in case a machine check >> exception is encoutered during the memcpy. >> >> This patch largely borrows from the copyuser_power7 >> logic and does not add the VMX optimizations, largely >> to keep the patch simple. If needed those optimizations >> can be folded in. > > So memcpy_mcsafe doesn't return number of bytes copied? > Huh, well that makes it simple. Well, not in current kernels, but we need to add that support or remove the direct call to copy_to_iter() in fs/dax.c. I'm looking right now to add "bytes remaining" support to the x86 memcpy_mcsafe(), but for copy_to_user we also need to handle bytes remaining for write faults. That fix is hopefully something that can land in an early 4.17-rc, but it won't be ready for -rc1.
Re: [RESEND 2/3] powerpc/memcpy: Add memcpy_mcsafe for pmem
On Thu, 5 Apr 2018 09:19:42 +1000 Balbir Singhwrote: > The pmem infrastructure uses memcpy_mcsafe in the pmem > layer so as to convert machine check excpetions into > a return value on failure in case a machine check > exception is encoutered during the memcpy. > > This patch largely borrows from the copyuser_power7 > logic and does not add the VMX optimizations, largely > to keep the patch simple. If needed those optimizations > can be folded in. So memcpy_mcsafe doesn't return number of bytes copied? Huh, well that makes it simple. Would be nice if there was an easy way to share this with the regular memcpy code... that's probably for another day though, probably better to let this settle down first. I didn't review exact instructions, but the approach looks right to me. Acked-by: Nicholas Piggin > > Signed-off-by: Balbir Singh > --- > arch/powerpc/include/asm/string.h | 2 + > arch/powerpc/lib/Makefile | 2 +- > arch/powerpc/lib/memcpy_mcsafe_64.S | 212 > > 3 files changed, 215 insertions(+), 1 deletion(-) > create mode 100644 arch/powerpc/lib/memcpy_mcsafe_64.S > > diff --git a/arch/powerpc/include/asm/string.h > b/arch/powerpc/include/asm/string.h > index 9b8cedf618f4..b7e872a64726 100644 > --- a/arch/powerpc/include/asm/string.h > +++ b/arch/powerpc/include/asm/string.h > @@ -30,7 +30,9 @@ extern void * memcpy_flushcache(void *,const void > *,__kernel_size_t); > #ifdef CONFIG_PPC64 > #define __HAVE_ARCH_MEMSET32 > #define __HAVE_ARCH_MEMSET64 > +#define __HAVE_ARCH_MEMCPY_MCSAFE > > +extern int memcpy_mcsafe(void *dst, const void *src, __kernel_size_t sz); > extern void *__memset16(uint16_t *, uint16_t v, __kernel_size_t); > extern void *__memset32(uint32_t *, uint32_t v, __kernel_size_t); > extern void *__memset64(uint64_t *, uint64_t v, __kernel_size_t); > diff --git a/arch/powerpc/lib/Makefile b/arch/powerpc/lib/Makefile > index 3c29c9009bbf..048afee9f518 100644 > --- a/arch/powerpc/lib/Makefile > +++ b/arch/powerpc/lib/Makefile > @@ -24,7 +24,7 @@ endif > > obj64-y += copypage_64.o copyuser_64.o mem_64.o hweight_64.o \ > copyuser_power7.o string_64.o copypage_power7.o memcpy_power7.o \ > -memcpy_64.o memcmp_64.o pmem.o > +memcpy_64.o memcmp_64.o pmem.o memcpy_mcsafe_64.o > > obj64-$(CONFIG_SMP) += locks.o > obj64-$(CONFIG_ALTIVEC) += vmx-helper.o > diff --git a/arch/powerpc/lib/memcpy_mcsafe_64.S > b/arch/powerpc/lib/memcpy_mcsafe_64.S > new file mode 100644 > index ..e7eaa9b6cded > --- /dev/null > +++ b/arch/powerpc/lib/memcpy_mcsafe_64.S > @@ -0,0 +1,212 @@ > +/* SPDX-License-Identifier: GPL-2.0 */ > +/* > + * Copyright (C) IBM Corporation, 2011 > + * Derived from copyuser_power7.s by Anton Blanchard > + * Author - Balbir Singh > + */ > +#include > +#include > + > + .macro err1 > +100: > + EX_TABLE(100b,.Ldo_err1) > + .endm > + > + .macro err2 > +200: > + EX_TABLE(200b,.Ldo_err2) > + .endm > + > +.Ldo_err2: > + ld r22,STK_REG(R22)(r1) > + ld r21,STK_REG(R21)(r1) > + ld r20,STK_REG(R20)(r1) > + ld r19,STK_REG(R19)(r1) > + ld r18,STK_REG(R18)(r1) > + ld r17,STK_REG(R17)(r1) > + ld r16,STK_REG(R16)(r1) > + ld r15,STK_REG(R15)(r1) > + ld r14,STK_REG(R14)(r1) > + addir1,r1,STACKFRAMESIZE > +.Ldo_err1: > + li r3,-EFAULT > + blr > + > + > +_GLOBAL(memcpy_mcsafe) > + cmpldi r5,16 > + blt .Lshort_copy > + > +.Lcopy: > + /* Get the source 8B aligned */ > + neg r6,r4 > + mtocrf 0x01,r6 > + clrldi r6,r6,(64-3) > + > + bf cr7*4+3,1f > +err1;lbz r0,0(r4) > + addir4,r4,1 > +err1;stb r0,0(r3) > + addir3,r3,1 > + > +1: bf cr7*4+2,2f > +err1;lhz r0,0(r4) > + addir4,r4,2 > +err1;sth r0,0(r3) > + addir3,r3,2 > + > +2: bf cr7*4+1,3f > +err1;lwz r0,0(r4) > + addir4,r4,4 > +err1;stw r0,0(r3) > + addir3,r3,4 > + > +3: sub r5,r5,r6 > + cmpldi r5,128 > + blt 5f > + > + mflrr0 > + stdur1,-STACKFRAMESIZE(r1) > + std r14,STK_REG(R14)(r1) > + std r15,STK_REG(R15)(r1) > + std r16,STK_REG(R16)(r1) > + std r17,STK_REG(R17)(r1) > + std r18,STK_REG(R18)(r1) > + std r19,STK_REG(R19)(r1) > + std r20,STK_REG(R20)(r1) > + std r21,STK_REG(R21)(r1) > + std r22,STK_REG(R22)(r1) > + std r0,STACKFRAMESIZE+16(r1) > + > + srdir6,r5,7 > + mtctr r6 > + > + /* Now do cacheline (128B) sized loads and stores. */ > + .align 5 > +4: > +err2;ld r0,0(r4) > +err2;ld r6,8(r4) > +err2;ld r7,16(r4) > +err2;ld r8,24(r4) >
[RESEND 2/3] powerpc/memcpy: Add memcpy_mcsafe for pmem
The pmem infrastructure uses memcpy_mcsafe in the pmem layer so as to convert machine check excpetions into a return value on failure in case a machine check exception is encoutered during the memcpy. This patch largely borrows from the copyuser_power7 logic and does not add the VMX optimizations, largely to keep the patch simple. If needed those optimizations can be folded in. Signed-off-by: Balbir Singh--- arch/powerpc/include/asm/string.h | 2 + arch/powerpc/lib/Makefile | 2 +- arch/powerpc/lib/memcpy_mcsafe_64.S | 212 3 files changed, 215 insertions(+), 1 deletion(-) create mode 100644 arch/powerpc/lib/memcpy_mcsafe_64.S diff --git a/arch/powerpc/include/asm/string.h b/arch/powerpc/include/asm/string.h index 9b8cedf618f4..b7e872a64726 100644 --- a/arch/powerpc/include/asm/string.h +++ b/arch/powerpc/include/asm/string.h @@ -30,7 +30,9 @@ extern void * memcpy_flushcache(void *,const void *,__kernel_size_t); #ifdef CONFIG_PPC64 #define __HAVE_ARCH_MEMSET32 #define __HAVE_ARCH_MEMSET64 +#define __HAVE_ARCH_MEMCPY_MCSAFE +extern int memcpy_mcsafe(void *dst, const void *src, __kernel_size_t sz); extern void *__memset16(uint16_t *, uint16_t v, __kernel_size_t); extern void *__memset32(uint32_t *, uint32_t v, __kernel_size_t); extern void *__memset64(uint64_t *, uint64_t v, __kernel_size_t); diff --git a/arch/powerpc/lib/Makefile b/arch/powerpc/lib/Makefile index 3c29c9009bbf..048afee9f518 100644 --- a/arch/powerpc/lib/Makefile +++ b/arch/powerpc/lib/Makefile @@ -24,7 +24,7 @@ endif obj64-y+= copypage_64.o copyuser_64.o mem_64.o hweight_64.o \ copyuser_power7.o string_64.o copypage_power7.o memcpy_power7.o \ - memcpy_64.o memcmp_64.o pmem.o + memcpy_64.o memcmp_64.o pmem.o memcpy_mcsafe_64.o obj64-$(CONFIG_SMP)+= locks.o obj64-$(CONFIG_ALTIVEC)+= vmx-helper.o diff --git a/arch/powerpc/lib/memcpy_mcsafe_64.S b/arch/powerpc/lib/memcpy_mcsafe_64.S new file mode 100644 index ..e7eaa9b6cded --- /dev/null +++ b/arch/powerpc/lib/memcpy_mcsafe_64.S @@ -0,0 +1,212 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* + * Copyright (C) IBM Corporation, 2011 + * Derived from copyuser_power7.s by Anton Blanchard + * Author - Balbir Singh + */ +#include +#include + + .macro err1 +100: + EX_TABLE(100b,.Ldo_err1) + .endm + + .macro err2 +200: + EX_TABLE(200b,.Ldo_err2) + .endm + +.Ldo_err2: + ld r22,STK_REG(R22)(r1) + ld r21,STK_REG(R21)(r1) + ld r20,STK_REG(R20)(r1) + ld r19,STK_REG(R19)(r1) + ld r18,STK_REG(R18)(r1) + ld r17,STK_REG(R17)(r1) + ld r16,STK_REG(R16)(r1) + ld r15,STK_REG(R15)(r1) + ld r14,STK_REG(R14)(r1) + addir1,r1,STACKFRAMESIZE +.Ldo_err1: + li r3,-EFAULT + blr + + +_GLOBAL(memcpy_mcsafe) + cmpldi r5,16 + blt .Lshort_copy + +.Lcopy: + /* Get the source 8B aligned */ + neg r6,r4 + mtocrf 0x01,r6 + clrldi r6,r6,(64-3) + + bf cr7*4+3,1f +err1; lbz r0,0(r4) + addir4,r4,1 +err1; stb r0,0(r3) + addir3,r3,1 + +1: bf cr7*4+2,2f +err1; lhz r0,0(r4) + addir4,r4,2 +err1; sth r0,0(r3) + addir3,r3,2 + +2: bf cr7*4+1,3f +err1; lwz r0,0(r4) + addir4,r4,4 +err1; stw r0,0(r3) + addir3,r3,4 + +3: sub r5,r5,r6 + cmpldi r5,128 + blt 5f + + mflrr0 + stdur1,-STACKFRAMESIZE(r1) + std r14,STK_REG(R14)(r1) + std r15,STK_REG(R15)(r1) + std r16,STK_REG(R16)(r1) + std r17,STK_REG(R17)(r1) + std r18,STK_REG(R18)(r1) + std r19,STK_REG(R19)(r1) + std r20,STK_REG(R20)(r1) + std r21,STK_REG(R21)(r1) + std r22,STK_REG(R22)(r1) + std r0,STACKFRAMESIZE+16(r1) + + srdir6,r5,7 + mtctr r6 + + /* Now do cacheline (128B) sized loads and stores. */ + .align 5 +4: +err2; ld r0,0(r4) +err2; ld r6,8(r4) +err2; ld r7,16(r4) +err2; ld r8,24(r4) +err2; ld r9,32(r4) +err2; ld r10,40(r4) +err2; ld r11,48(r4) +err2; ld r12,56(r4) +err2; ld r14,64(r4) +err2; ld r15,72(r4) +err2; ld r16,80(r4) +err2; ld r17,88(r4) +err2; ld r18,96(r4) +err2; ld r19,104(r4) +err2; ld r20,112(r4) +err2; ld r21,120(r4) + addir4,r4,128 +err2; std r0,0(r3) +err2; std r6,8(r3) +err2; std r7,16(r3) +err2; std r8,24(r3) +err2; std r9,32(r3) +err2; std r10,40(r3) +err2; std r11,48(r3) +err2; std r12,56(r3) +err2; std r14,64(r3) +err2; std r15,72(r3) +err2; std r16,80(r3) +err2; std r17,88(r3) +err2; std r18,96(r3) +err2;