ARM: V7M: align v7m_dma_inv_range() with v7 counterpart
PATCH 1/1 ARM: V7M: align v7m_dma_inv_range() with v7 counterpart
Chris has discovered and reported that v7_dma_inv_range() may corrupt memory if address range is not aligned to cache line size.
Since the whole cache-v7m.S was lifted form cache-v7.S the same observation applies to v7m_dma_inv_range(). So the fix just mirrors what has been done for v7 with a little specific of M-class.
--- arch/arm/mm/cache-v7m.S | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-)
arch/arm/mm/cache-v7m.S +8 -5
@@ -73,9 +73,11 @@ /* * dcimvac: Invalidate data cache line by MVA to PoC */ -.macro dcimvac, rt, tmp - v7m_cacheop \rt, \tmp, V7M_SCB_DCIMVAC +.irp c,,eq,ne,cs,cc,mi,pl,vs,vc,hi,ls,ge,lt,gt,le,hs,lo +.macro dcimvac\c, rt, tmp + v7m_cacheop \rt, \tmp, V7M_SCB_DCIMVAC, \c .endm +.endr /* * dccmvau: Clean data cache line by MVA to PoU
@@ -369,14 +371,15 @@ v7m_dma_inv_range: tst r0, r3 bic r0, r0, r3 dccimvacne r0, r3 + addne r0, r0, r2 subne r3, r2, #1 @ restore r3, corrupted by v7m's dccimvac tst r1, r3 bic r1, r1, r3 dccimvacne r1, r3 1: - dcimvac r0, r3 - add r0, r0, r2 - cmp r0, r1 + dcimvaclo r0, r3 + addlo r0, r0, r2 + cmplo r0, r1 blo 1b dsb st ret lr
Signature · 1 line
1.9.1
PATCH 1/1 ARM: V7M: align v7m_dma_inv_range() with v7 counterpart
Hi Vladimir,
Quoted 52 lines · Vladimir Murzin
Chris has discovered and reported that
v7_dma_inv_range()may corrupt memory if address range is not aligned to cache line size.Since the whole
cache-v7m.Swas lifted formcache-v7.Sthe same observation applies tov7m_dma_inv_range(). So the fix just mirrors what has been done for v7 with a little specific of M-class.Cc Chris Cole <[email protected]>Signed-off-by Vladimir Murzin <[email protected]>--- arch/arm/mm/cache-v7m.S | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-)arch/arm/mm/cache-v7m.S +8
-6@@ -73,9 +73,11 @@ /* * dcimvac: Invalidate data cache line by MVA to PoC */ -.macro dcimvac, rt, tmp - v7m_cacheop \rt, \tmp, V7M_SCB_DCIMVAC +.irp c,,eq,ne,cs,cc,mi,pl,vs,vc,hi,ls,ge,lt,gt,le,hs,lo +.macro dcimvac\c, rt, tmp + v7m_cacheop \rt, \tmp, V7M_SCB_DCIMVAC, \c .endm +.endr /* * dccmvau: Clean data cache line by MVA to PoU@@ -369,14 +371,15 @@ v7m_dma_inv_range: tst r0, r3 bic r0, r0, r3 dccimvacne r0, r3 + addne r0, r0, r2 subne r3, r2, #1 @ restore r3, corrupted by v7m's dccimvac tst r1, r3 bic r1, r1, r3 dccimvacne r1, r3 1: - dcimvac r0, r3 - add r0, r0, r2 - cmp r0, r1 + dcimvaclo r0, r3 + addlo r0, r0, r2 + cmplo r0, r1 blo 1b dsb st ret lr --
Quoted 2 lines
1.9.1
There needs to be "cmp r0, r1" added after "1:" label.
Thanks,
Chris
PATCH 1/1 ARM: V7M: align v7m_dma_inv_range() with v7 counterpart
Hi Chris,
Quoted 51 lines · Chris Cole
Hi Vladimir,
Quoted 48 lines · Vladimir Murzin
Chris has discovered and reported that
v7_dma_inv_range()may corrupt memory if address range is not aligned to cache line size.Since the whole
cache-v7m.Swas lifted formcache-v7.Sthe same observation applies tov7m_dma_inv_range(). So the fix just mirrors what has been done for v7 with a little specific of M-class.Cc Chris Cole <[email protected]>Signed-off-by Vladimir Murzin <[email protected]>--- arch/arm/mm/cache-v7m.S | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-)arch/arm/mm/cache-v7m.S +8
-5@@ -73,9 +73,11 @@ /* * dcimvac: Invalidate data cache line by MVA to PoC */ -.macro dcimvac, rt, tmp - v7m_cacheop \rt, \tmp, V7M_SCB_DCIMVAC +.irp c,,eq,ne,cs,cc,mi,pl,vs,vc,hi,ls,ge,lt,gt,le,hs,lo +.macro dcimvac\c, rt, tmp + v7m_cacheop \rt, \tmp, V7M_SCB_DCIMVAC, \c .endm +.endr /* * dccmvau: Clean data cache line by MVA to PoU@@ -369,14 +371,15 @@ v7m_dma_inv_range: tst r0, r3 bic r0, r0, r3 dccimvacne r0, r3 + addne r0, r0, r2 subne r3, r2, #1 @ restore r3, corrupted by v7m's dccimvac tst r1, r3 bic r1, r1, r3 dccimvacne r1, r3 1: - dcimvac r0, r3 - add r0, r0, r2 - cmp r0, r1 + dcimvaclo r0, r3 + addlo r0, r0, r2 + cmplo r0, r1
Quoted 8 lines
Quoted 6 lines
arch/arm/mm/cache-v7m.S +0
-1blo 1b dsb st ret lr --1.9.1
There needs to be "cmp r0, r1" added after "1:" label.
Nice catch! I'll send update shortly.
Thanks
Vladimir
Quoted 4 lines
Thanks,
Chris
Russell King - ARM Linux
maintainerPATCH 1/1 ARM: V7M: align v7m_dma_inv_range() with v7 counterpart
Quoted 63 lines · Vladimir Murzin
Hi Chris,
Quoted 51 lines · Chris Cole
Hi Vladimir,
Quoted 48 lines · Vladimir Murzin
Chris has discovered and reported that
v7_dma_inv_range()may corrupt memory if address range is not aligned to cache line size.Since the whole
cache-v7m.Swas lifted formcache-v7.Sthe same observation applies tov7m_dma_inv_range(). So the fix just mirrors what has been done for v7 with a little specific of M-class.Cc Chris Cole <[email protected]>Signed-off-by Vladimir Murzin <[email protected]>--- arch/arm/mm/cache-v7m.S | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-)arch/arm/mm/cache-v7m.S +8
-5@@ -73,9 +73,11 @@ /* * dcimvac: Invalidate data cache line by MVA to PoC */ -.macro dcimvac, rt, tmp - v7m_cacheop \rt, \tmp, V7M_SCB_DCIMVAC +.irp c,,eq,ne,cs,cc,mi,pl,vs,vc,hi,ls,ge,lt,gt,le,hs,lo +.macro dcimvac\c, rt, tmp + v7m_cacheop \rt, \tmp, V7M_SCB_DCIMVAC, \c .endm +.endr /* * dccmvau: Clean data cache line by MVA to PoU@@ -369,14 +371,15 @@ v7m_dma_inv_range: tst r0, r3 bic r0, r0, r3 dccimvacne r0, r3 + addne r0, r0, r2 subne r3, r2, #1 @ restore r3, corrupted by v7m's dccimvac tst r1, r3 bic r1, r1, r3 dccimvacne r1, r3 1: - dcimvac r0, r3 - add r0, r0, r2 - cmp r0, r1 + dcimvaclo r0, r3 + addlo r0, r0, r2 + cmplo r0, r1Quoted 8 lines
Quoted 6 lines
arch/arm/mm/cache-v7m.S +0
-1blo 1b dsb st ret lr --1.9.1
There needs to be "cmp r0, r1" added after "1:" label.
Quoted 1 line
Nice catch! I'll send update shortly.
No, actually, the cmp needs to be _before_ the label, as per the v7 version.
Signature · 3 lines
RMK's Patch system: http://www.armlinux.org.uk/developer/patches/ FTTC broadband for 0.8mile line in suburbia: sync at 12.1Mbps down 622kbps up According to speedtest.net: 11.9Mbps down 500kbps up
PATCH 1/1 ARM: V7M: align v7m_dma_inv_range() with v7 counterpart
Quoted 64 lines · Russell King - ARM Linux
Quoted 63 lines · Vladimir Murzin
Hi Chris,
Quoted 59 lines · Chris Cole
Hi Vladimir,
Quoted 52 lines · Vladimir Murzin
Chris has discovered and reported that
v7_dma_inv_range()may corrupt memory if address range is not aligned to cache line size.Since the whole
cache-v7m.Swas lifted formcache-v7.Sthe same observation applies tov7m_dma_inv_range(). So the fix just mirrors what has been done for v7 with a little specific of M-class.Cc Chris Cole <[email protected]>Signed-off-by Vladimir Murzin <[email protected]>--- arch/arm/mm/cache-v7m.S | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-)arch/arm/mm/cache-v7m.S +8
-6@@ -73,9 +73,11 @@ /* * dcimvac: Invalidate data cache line by MVA to PoC */ -.macro dcimvac, rt, tmp - v7m_cacheop \rt, \tmp, V7M_SCB_DCIMVAC +.irp c,,eq,ne,cs,cc,mi,pl,vs,vc,hi,ls,ge,lt,gt,le,hs,lo +.macro dcimvac\c, rt, tmp + v7m_cacheop \rt, \tmp, V7M_SCB_DCIMVAC, \c .endm +.endr /* * dccmvau: Clean data cache line by MVA to PoU@@ -369,14 +371,15 @@ v7m_dma_inv_range: tst r0, r3 bic r0, r0, r3 dccimvacne r0, r3 + addne r0, r0, r2 subne r3, r2, #1 @ restore r3, corrupted by v7m's dccimvac tst r1, r3 bic r1, r1, r3 dccimvacne r1, r3 1: - dcimvac r0, r3 - add r0, r0, r2 - cmp r0, r1 + dcimvaclo r0, r3 + addlo r0, r0, r2 + cmplo r0, r1 blo 1b dsb st ret lr --Quoted 2 lines
1.9.1
There needs to be "cmp r0, r1" added after "1:" label.
Quoted 5 lines
Quoted 1 line
Nice catch! I'll send update shortly.
No, actually, the cmp needs to be _before_ the label, as per the v7 version.
Yup, I also noticed that, so v2 already have cmp before the label.
Since you in thread, are you happy if both patches go into your patch system?
Thanks
Vladimir