Skip to contentSkip to conversation

raid6/altivec: Add vpermxor implementation for raid6 Q syndrome

PATCH In review Matt Brown 3 participants 6 messages Aug 4, 2017 – Mar 20, 2018 MD RAID

lore
Share & tools
lore

Matt Brown

author
#0 lore raw

PATCH v6 1/2 raid6/altivec: Add vpermxor implementation for raid6 Q syndrome

This patch uses the vpermxor instruction to optimise the raid6 Q syndrome. This instruction was made available with POWER8, ISA version 2.07. It allows for both vperm and vxor instructions to be done in a single instruction. This has been tested for correctness on a ppc64le vm with a basic RAID6 setup containing 5 drives.

The performance benchmarks are from the raid6test in the /lib/raid6/test directory. These results are from an IBM Firestone machine with ppc64le architecture. The benchmark results show a 35% speed increase over the best existing algorithm for powerpc (altivec). The raid6test has also been run on a big-endian ppc64 vm to ensure it also works for big-endian architectures.

Performance benchmarks:

		raid6: altivecx4 gen() 18773 MB/s
		raid6: altivecx8 gen() 19438 MB/s

		raid6: vpermxor4 gen() 25112 MB/s
		raid6: vpermxor8 gen() 26279 MB/s

---
v6:
- added vpermxor files to .gitignore
- fixup whitespace
- added vpermxor objs to test/Makefile
v5:
- moved altivec.uc fix into other patch in series

---
 include/linux/raid/pq.h |   4 ++
 lib/raid6/.gitignore    |   1 +
 lib/raid6/Makefile      |  27 ++++++++++++-
 lib/raid6/algos.c       |   4 ++
 lib/raid6/test/Makefile |  17 +++++++-
 lib/raid6/vpermxor.uc   | 104 ++++++++++++++++++++++++++++++++++++++++++++++++
 6 files changed, 154 insertions(+), 3 deletions(-)
 create mode 100644 lib/raid6/vpermxor.uc
include/linux/raid/pq.h +4 -0
index 4d57bba..3df9aa6 100644
@@ -107,6 +107,10 @@ extern const struct raid6_calls raid6_avx512x2;
 extern const struct raid6_calls raid6_avx512x4;
 extern const struct raid6_calls raid6_tilegx8;
 extern const struct raid6_calls raid6_s390vx8;
+extern const struct raid6_calls raid6_vpermxor1;
+extern const struct raid6_calls raid6_vpermxor2;
+extern const struct raid6_calls raid6_vpermxor4;
+extern const struct raid6_calls raid6_vpermxor8;
 
 struct raid6_recov_calls {
 	void (*data2)(int, size_t, int, int, void **);
lib/raid6/.gitignore +1 -0
index f01b1cb..3de0d89 100644
@@ -4,3 +4,4 @@ int*.c
 tables.c
 neon?.c
 s390vx?.c
+vpermxor*.c
lib/raid6/Makefile +26 -1
index 3057011..db095a7 100644
@@ -4,7 +4,8 @@ raid6_pq-y	+= algos.o recov.o tables.o int1.o int2.o int4.o \
 		   int8.o int16.o int32.o
 
 raid6_pq-$(CONFIG_X86) += recov_ssse3.o recov_avx2.o mmx.o sse1.o sse2.o avx2.o avx512.o recov_avx512.o
-raid6_pq-$(CONFIG_ALTIVEC) += altivec1.o altivec2.o altivec4.o altivec8.o
+raid6_pq-$(CONFIG_ALTIVEC) += altivec1.o altivec2.o altivec4.o altivec8.o \
+                              vpermxor1.o vpermxor2.o vpermxor4.o vpermxor8.o
 raid6_pq-$(CONFIG_KERNEL_MODE_NEON) += neon.o neon1.o neon2.o neon4.o neon8.o
 raid6_pq-$(CONFIG_TILEGX) += tilegx8.o
 raid6_pq-$(CONFIG_S390) += s390vx8.o recov_s390xc.o
@@ -88,6 +89,30 @@ $(obj)/altivec8.c:   UNROLL := 8
 $(obj)/altivec8.c:   $(src)/altivec.uc $(src)/unroll.awk FORCE
 	$(call if_changed,unroll)
 
+CFLAGS_vpermxor1.o += $(altivec_flags)
+targets += vpermxor1.c
+$(obj)/vpermxor1.c: UNROLL := 1
+$(obj)/vpermxor1.c: $(src)/vpermxor.uc $(src)/unroll.awk FORCE
+	$(call if_changed,unroll)
+
+CFLAGS_vpermxor2.o += $(altivec_flags)
+targets += vpermxor2.c
+$(obj)/vpermxor2.c: UNROLL := 2
+$(obj)/vpermxor2.c: $(src)/vpermxor.uc $(src)/unroll.awk FORCE
+	$(call if_changed,unroll)
+
+CFLAGS_vpermxor4.o += $(altivec_flags)
+targets += vpermxor4.c
+$(obj)/vpermxor4.c: UNROLL := 4
+$(obj)/vpermxor4.c: $(src)/vpermxor.uc $(src)/unroll.awk FORCE
+	$(call if_changed,unroll)
+
+CFLAGS_vpermxor8.o += $(altivec_flags)
+targets += vpermxor8.c
+$(obj)/vpermxor8.c: UNROLL := 8
+$(obj)/vpermxor8.c: $(src)/vpermxor.uc $(src)/unroll.awk FORCE
+	$(call if_changed,unroll)
+
 CFLAGS_neon1.o += $(NEON_FLAGS)
 targets += neon1.c
 $(obj)/neon1.c:   UNROLL := 1
lib/raid6/algos.c +4 -0
index 7857049..edd4f69 100644
@@ -74,6 +74,10 @@ const struct raid6_calls * const raid6_algos[] = {
 	&raid6_altivec2,
 	&raid6_altivec4,
 	&raid6_altivec8,
+	&raid6_vpermxor1,
+	&raid6_vpermxor2,
+	&raid6_vpermxor4,
+	&raid6_vpermxor8,
 #endif
 #if defined(CONFIG_TILEGX)
 	&raid6_tilegx8,
lib/raid6/test/Makefile +15 -2
index 2c7b60e..a14be53 100644
@@ -47,7 +47,8 @@ else
                          gcc -c -x c - >&/dev/null && \
                          rm ./-.o && echo yes)
         ifeq ($(HAS_ALTIVEC),yes)
-                OBJS += altivec1.o altivec2.o altivec4.o altivec8.o
+                OBJS += altivec1.o altivec2.o altivec4.o altivec8.o \
+                        vpermxor1.o vpermxor2.o vpermxor4.o vpermxor8.o
         endif
 endif
 ifeq ($(ARCH),tilegx)
@@ -97,6 +98,18 @@ altivec4.c: altivec.uc ../unroll.awk
 altivec8.c: altivec.uc ../unroll.awk
 	$(AWK) ../unroll.awk -vN=8 < altivec.uc > $@
 
+vpermxor1.c: vpermxor.uc ../unroll.awk
+	$(AWK) ../unroll.awk -vN=1 < vpermxor.uc > $@
+
+vpermxor2.c: vpermxor.uc ../unroll.awk
+	$(AWK) ../unroll.awk -vN=2 < vpermxor.uc > $@
+
+vpermxor4.c: vpermxor.uc ../unroll.awk
+	$(AWK) ../unroll.awk -vN=4 < vpermxor.uc > $@
+
+vpermxor8.c: vpermxor.uc ../unroll.awk
+	$(AWK) ../unroll.awk -vN=8 < vpermxor.uc > $@
+
 int1.c: int.uc ../unroll.awk
 	$(AWK) ../unroll.awk -vN=1 < int.uc > $@
 
@@ -122,7 +135,7 @@ tables.c: mktables
 	./mktables > tables.c
 
 clean:
-	rm -f *.o *.a mktables mktables.c *.uc int*.c altivec*.c neon*.c tables.c raid6test
+	rm -f *.o *.a mktables mktables.c *.uc int*.c altivec*.c vpermxor*.c neon*.c tables.c raid6test
 	rm -f tilegx*.c
 
 spotless: clean
lib/raid6/vpermxor.uc new file +104 -0

104 lines not on this page · raw message

Signature · 1 line
2.9.3
author
#1 lore raw

PATCH v6 2/2 lib/raid6: Build proper raid6test files on powerpc

Previously the raid6 test Makefile did not build the POWER specific files (altivec and vpermxor).
This patch fixes the bug, so that all appropriate files for powerpc are built.

This patch also fixes the missing and mismatched ifdef statements to allow the altivec.uc file to be built correctly.

---
v6:
- remove vpermxor objs from this patch
v5:
- moved altivec.uc fix into this patch

---
 lib/raid6/altivec.uc    | 3 +++
 lib/raid6/test/Makefile | 5 +++--
 2 files changed, 6 insertions(+), 2 deletions(-)
lib/raid6/altivec.uc +3 -0
index 682aae8..d20ed0d 100644
@@ -24,10 +24,13 @@
 
 #include <linux/raid/pq.h>
 
+#ifdef CONFIG_ALTIVEC
+
 #include <altivec.h>
 #ifdef __KERNEL__
 # include <asm/cputable.h>
 # include <asm/switch_to.h>
+#endif /* __KERNEL__ */
 
 /*
  * This is the C data type to use.  We use a vector of
lib/raid6/test/Makefile +3 -2
index a14be53..b64a267 100644
@@ -44,9 +44,10 @@ else ifeq ($(HAS_NEON),yes)
         CFLAGS += -DCONFIG_KERNEL_MODE_NEON=1
 else
         HAS_ALTIVEC := $(shell printf '\#include <altivec.h>\nvector int a;\n' |\
-                         gcc -c -x c - >&/dev/null && \
-                         rm ./-.o && echo yes)
+                         gcc -c -x c - >/dev/null && rm ./-.o && echo yes)
         ifeq ($(HAS_ALTIVEC),yes)
+                CFLAGS += -I../../../arch/powerpc/include
+                CFLAGS += -DCONFIG_ALTIVEC
                 OBJS += altivec1.o altivec2.o altivec4.o altivec8.o \
                         vpermxor1.o vpermxor2.o vpermxor4.o vpermxor8.o
         endif
Signature · 1 line
2.9.3
#2 lore raw
In reply to Matt Brown
Quoted 29 lines · Matt Brown

This patch uses the vpermxor instruction to optimise the raid6 Q syndrome. This instruction was made available with POWER8, ISA version 2.07. It allows for both vperm and vxor instructions to be done in a single instruction. This has been tested for correctness on a ppc64le vm with a basic RAID6 setup containing 5 drives.

The performance benchmarks are from the raid6test in the /lib/raid6/test directory. These results are from an IBM Firestone machine with ppc64le architecture. The benchmark results show a 35% speed increase over the best existing algorithm for powerpc (altivec). The raid6test has also been run on a big-endian ppc64 vm to ensure it also works for big-endian architectures.

Performance benchmarks:

		raid6: altivecx4 gen() 18773 MB/s
		raid6: altivecx8 gen() 19438 MB/s

		raid6: vpermxor4 gen() 25112 MB/s
		raid6: vpermxor8 gen() 26279 MB/s

---
v6:
- added vpermxor files to .gitignore
- fixup whitespace
- added vpermxor objs to test/Makefile
v5:
- moved altivec.uc fix into other patch in series

Quoted 9 lines
---
 include/linux/raid/pq.h |   4 ++
 lib/raid6/.gitignore    |   1 +
 lib/raid6/Makefile      |  27 ++++++++++++-
 lib/raid6/algos.c       |   4 ++
 lib/raid6/test/Makefile |  17 +++++++-
 lib/raid6/vpermxor.uc   | 104 ++++++++++++++++++++++++++++++++++++++++++++++++
 6 files changed, 154 insertions(+), 3 deletions(-)
 create mode 100644 lib/raid6/vpermxor.uc

This version at least is not Cc'ed to any of the folks that get_maintainers.pl identifies for these files:

$ ./scripts/get_maintainer.pl -f lib/raid6
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]

This seems like mostly a list of random folks who've touched this code, but maybe some of them would have comments?

cheers

author
#3 lore raw
Quoted 54 lines · Michael Ellerman
Quoted 29 lines · Matt Brown

This patch uses the vpermxor instruction to optimise the raid6 Q syndrome. This instruction was made available with POWER8, ISA version 2.07. It allows for both vperm and vxor instructions to be done in a single instruction. This has been tested for correctness on a ppc64le vm with a basic RAID6 setup containing 5 drives.

The performance benchmarks are from the raid6test in the /lib/raid6/test directory. These results are from an IBM Firestone machine with ppc64le architecture. The benchmark results show a 35% speed increase over the best existing algorithm for powerpc (altivec). The raid6test has also been run on a big-endian ppc64 vm to ensure it also works for big-endian architectures.

Performance benchmarks:

              raid6: altivecx4 gen() 18773 MB/s
              raid6: altivecx8 gen() 19438 MB/s

              raid6: vpermxor4 gen() 25112 MB/s
              raid6: vpermxor8 gen() 26279 MB/s

---
v6:
- added vpermxor files to .gitignore
- fixup whitespace
- added vpermxor objs to test/Makefile
v5:
- moved altivec.uc fix into other patch in series

Quoted 9 lines
---
 include/linux/raid/pq.h |   4 ++
 lib/raid6/.gitignore    |   1 +
 lib/raid6/Makefile      |  27 ++++++++++++-
 lib/raid6/algos.c       |   4 ++
 lib/raid6/test/Makefile |  17 +++++++-
 lib/raid6/vpermxor.uc   | 104 ++++++++++++++++++++++++++++++++++++++++++++++++
 6 files changed, 154 insertions(+), 3 deletions(-)
 create mode 100644 lib/raid6/vpermxor.uc

This version at least is not Cc'ed to any of the folks that get_maintainers.pl identifies for these files:

$ ./scripts/get_maintainer.pl -f lib/raid6
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
[email protected]
Quoted 3 lines

This seems like mostly a list of random folks who've touched this code, but maybe some of them would have comments?

Ah my bad. I've CC'ed them into this email chain. Apologies for not including you guys in the original email. Here is a link to the patchworks patch:
http://patchwork.ozlabs.org/patch/797576/

Thanks,
Matt Brown

#4 lore raw
In reply to Matt Brown
Quoted 14 lines · Matt Brown

This patch uses the vpermxor instruction to optimise the raid6 Q syndrome. This instruction was made available with POWER8, ISA version 2.07. It allows for both vperm and vxor instructions to be done in a single instruction. This has been tested for correctness on a ppc64le vm with a basic RAID6 setup containing 5 drives.

The performance benchmarks are from the raid6test in the /lib/raid6/test directory. These results are from an IBM Firestone machine with ppc64le architecture. The benchmark results show a 35% speed increase over the best existing algorithm for powerpc (altivec). The raid6test has also been run on a big-endian ppc64 vm to ensure it also works for big-endian architectures.

Performance benchmarks:

Quoted 8 lines
		raid6: altivecx4 gen() 18773 MB/s
		raid6: altivecx8 gen() 19438 MB/s

		raid6: vpermxor4 gen() 25112 MB/s
		raid6: vpermxor8 gen() 26279 MB/s

Series applied to powerpc next, thanks.

https://git.kernel.org/powerpc/c/2de95953c4e6ad54c9bee5e6a5518d

cheers

#5 lore raw
In reply to Matt Brown
Quoted 14 lines · Matt Brown

This patch uses the vpermxor instruction to optimise the raid6 Q syndrome. This instruction was made available with POWER8, ISA version 2.07. It allows for both vperm and vxor instructions to be done in a single instruction. This has been tested for correctness on a ppc64le vm with a basic RAID6 setup containing 5 drives.

The performance benchmarks are from the raid6test in the /lib/raid6/test directory. These results are from an IBM Firestone machine with ppc64le architecture. The benchmark results show a 35% speed increase over the best existing algorithm for powerpc (altivec). The raid6test has also been run on a big-endian ppc64 vm to ensure it also works for big-endian architectures.

Performance benchmarks:

Quoted 8 lines
		raid6: altivecx4 gen() 18773 MB/s
		raid6: altivecx8 gen() 19438 MB/s

		raid6: vpermxor4 gen() 25112 MB/s
		raid6: vpermxor8 gen() 26279 MB/s

Series applied to powerpc next, thanks.

https://git.kernel.org/powerpc/c/751ba79cc552c146595cd439b21c4f

cheers