dma: support DMA zone starting above 4GB
v3 cover dma: support DMA zone starting above 4GB
DMA zones code assumes that DMA lower limit is zero. When there is no RAM below 4GB, arm64 platform code sets DMA/DMA32 zone limits to cover the entire RAM[0].
My target platform has RAM starting at 32GB. Devices with 30-bit DMA mask are mapped to 1GB at the bottom of RAM, between 32GB - 33GB. DMA zone over the entire RAM breaks DMA allocation for these devices.
In response to a previous RFC hack[1] Catalin Marinas suggested to add a separate offset value as base address for the DMA zone, and then refined the suggestion to use start of RAM[3]. This RFC series attempts to implement that suggestion.
With this series applied, the DMA zone covers the right RAM range for my platform.
v3:
* Rebase on v6.11-rc1.
* Drop zone_dma_base. Use memblock_start_of_DRAM() instead.
* Drop DT patches. Low DMA range limit no longer needed.
* Add patch to improve dma_direct_optimal_gfp_mask() heuristics as Catalin suggested.
RFC v2:
* Add patch from Catalin[2] changing zone_dma_bits to zone_dma_limit to simplify subsequent patches
* Test on real hardware
RFC v1: https://lore.kernel.org/all/[email protected]/
[0] See commit 791ab8b2e3db ("arm64: Ignore any DMA offsets in the max_zone_phys() calculation")
[2] https://lore.kernel.org/all/[email protected]/
[3] https://lore.kernel.org/all/[email protected]/
Baruch Siach (2):
dma-mapping: improve DMA zone selection
dma-direct: use RAM start to offset zone_dma_limit
Catalin Marinas (1):
dma-mapping: replace zone_dma_bits by zone_dma_limit
arch/arm64/mm/init.c | 34 +++++++++++++--------------------- arch/powerpc/mm/mem.c | 9 ++++----- arch/s390/mm/init.c | 2 +- include/linux/dma-direct.h | 2 +- kernel/dma/direct.c | 15 ++++++++------- kernel/dma/pool.c | 3 ++- kernel/dma/swiotlb.c | 4 ++-- 7 files changed, 31 insertions(+), 38 deletions(-)
Signature · 1 line
2.43.0
PATCH v3 1/3 dma-mapping: improve DMA zone selection
When device DMA limit does not fit in DMA32 zone it should use DMA zone, even when DMA zone is stricter than needed.
Same goes for devices that can't allocate from the entire normal zone. Limit to DMA32 in that case.
--- kernel/dma/direct.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
kernel/dma/direct.c +3 -3
@@ -4,7 +4,7 @@* * DMA operations that map physical memory directly without using an IOMMU. */ -#include <linux/memblock.h> /* for max_pfn */ +#include <linux/memblock.h> #include <linux/export.h> #include <linux/mm.h> #include <linux/dma-map-ops.h>
@@ -59,9 +59,9 @@ static gfp_t dma_direct_optimal_gfp_mask(struct device *dev, u64 *phys_limit)* zones. */ *phys_limit =dma_to_phys (dev, dma_limit); -if (*phys_limit <=DMA_BIT_MASK (zone_dma_bits)) +if (*phys_limit <DMA_BIT_MASK (32 ))return GFP_DMA ; -if (*phys_limit <=DMA_BIT_MASK (32 )) +if (*phys_limit <memblock_end_of_DRAM ())return GFP_DMA32 ;return 0 ; }
Signature · 1 line
2.43.0
PATCH v3 2/3 dma-mapping: replace zone_dma_bits by zone_dma_limit
In reply to Baruch SiachFrom: Catalin Marinas <[email protected]>
Hardware DMA limit might not be power of 2. When RAM range starts above 0, say 4GB, DMA limit of 30 bits should end at 5GB. A single high bit can not encode this limit.
Use direct phys_addr_t limit address for DMA zone limit.
--- arch/arm64/mm/init.c | 34 +++++++++++++--------------------- arch/powerpc/mm/mem.c | 9 ++++----- arch/s390/mm/init.c | 2 +- include/linux/dma-direct.h | 2 +- kernel/dma/direct.c | 4 ++-- kernel/dma/pool.c | 2 +- kernel/dma/swiotlb.c | 4 ++-- 7 files changed, 24 insertions(+), 33 deletions(-)
arch/arm64/mm/init.c +13 -21
@@ -114,36 +114,28 @@ static void __init arch_reserve_crashkernel(void) low_size, high); } -/* -* Return the maximum physical address for a zone accessible by the given bits -* limit. If DRAM starts above 32-bit, expand the zone to the maximum -* available memory, otherwise cap it at 32-bit. -*/ -static phys_addr_t __init max_zone_phys (unsigned int zone_bits) +static phys_addr_t __init max_zone_phys (phys_addr_t zone_limit) { -phys_addr_t zone_mask =DMA_BIT_MASK (zone_bits); -phys_addr_t phys_start =memblock_start_of_DRAM (); - -if (phys_start >U32_MAX ) - zone_mask =PHYS_ADDR_MAX ; -else if (phys_start > zone_mask) - zone_mask =U32_MAX ; +/* We have RAM in low 32-bit area, keep DMA zone there */ +if (memblock_start_of_DRAM () <U32_MAX ) + zone_limit =min (U32_MAX , zone_limit); -return min (zone_mask,memblock_end_of_DRAM () -1 ) +1 ; +return min (zone_limit,memblock_end_of_DRAM () -1 ) +1 ; }static void __init zone_sizes_init (void ) {unsigned long max_zone_pfns[MAX_NR_ZONES ] = {0 }; -unsigned int __maybe_unused acpi_zone_dma_bits; -unsigned int __maybe_unused dt_zone_dma_bits; -phys_addr_t __maybe_unused dma32_phys_limit =max_zone_phys (32 ); +phys_addr_t __maybe_unused acpi_zone_dma_limit; +phys_addr_t __maybe_unused dt_zone_dma_limit; +phys_addr_t __maybe_unused dma32_phys_limit = +max_zone_phys (DMA_BIT_MASK (32 ));#ifdef CONFIG_ZONE_DMA - acpi_zone_dma_bits =fls64 (acpi_iort_dma_get_max_cpu_address ()); - dt_zone_dma_bits =fls64 (of_dma_get_max_cpu_address (NULL )); - zone_dma_bits =min3 (32U , dt_zone_dma_bits, acpi_zone_dma_bits); - arm64_dma_phys_limit =max_zone_phys (zone_dma_bits); + acpi_zone_dma_limit =acpi_iort_dma_get_max_cpu_address (); + dt_zone_dma_limit =of_dma_get_max_cpu_address (NULL ); + zone_dma_limit =min (dt_zone_dma_limit, acpi_zone_dma_limit); + arm64_dma_phys_limit =max_zone_phys (zone_dma_limit); max_zone_pfns[ZONE_DMA ] =PFN_DOWN (arm64_dma_phys_limit);#endif #ifdef CONFIG_ZONE_DMA32
arch/powerpc/mm/mem.c +4 -5
@@ -216,7 +216,7 @@ static int __init mark_nonram_nosave(void)* everything else. GFP_DMA32 page allocations automatically fall back to * ZONE_DMA. * -* By using 31-bit unconditionally, we can exploit zone_dma_bits to inform the +* By using 31-bit unconditionally, we can exploit zone_dma_limit to inform the * generic DMA mapping code. 32-bit only devices (if not handled by an IOMMU * anyway) will take a first dip into ZONE_NORMAL and get otherwise served by * ZONE_DMA.
@@ -252,13 +252,12 @@ void __init paging_init(void)* powerbooks. */ if (IS_ENABLED (CONFIG_PPC32 )) - zone_dma_bits =30 ; + zone_dma_limit =DMA_BIT_MASK (30 );else - zone_dma_bits =31 ; + zone_dma_limit =DMA_BIT_MASK (31 );#ifdef CONFIG_ZONE_DMA - max_zone_pfns[ZONE_DMA ] =min (max_low_pfn, -1UL << (zone_dma_bits -PAGE_SHIFT )); + max_zone_pfns[ZONE_DMA ] =min (max_low_pfn, zone_dma_limit >>PAGE_SHIFT );#endif max_zone_pfns[ZONE_NORMAL ] = max_low_pfn;#ifdef CONFIG_HIGHMEM
arch/s390/mm/init.c +1 -1
@@ -97,7 +97,7 @@ void __init paging_init(void)vmem_map_init ();sparse_init (); - zone_dma_bits =31 ; + zone_dma_limit =DMA_BIT_MASK (31 );memset (max_zone_pfns,0 ,sizeof (max_zone_pfns)); max_zone_pfns[ZONE_DMA ] =virt_to_pfn (MAX_DMA_ADDRESS ); max_zone_pfns[ZONE_NORMAL ] = max_low_pfn;
include/linux/dma-direct.h +1 -1
@@ -12,7 +12,7 @@#include <linux/mem_encrypt.h> #include <linux/swiotlb.h> -extern unsigned int zone_dma_bits; +extern phys_addr_t zone_dma_limit;/* * Record the mapping of CPU physical to DMA addresses for a given region.
kernel/dma/direct.c +2 -2
@@ -20,7 +20,7 @@* it for entirely different regions. In that case the arch code needs to * override the variable below for dma-direct to work properly. */ -unsigned int zone_dma_bits__ro_after_init =24 ; +phys_addr_t zone_dma_limit__ro_after_init =DMA_BIT_MASK (24 );static inline dma_addr_t phys_to_dma_direct (struct device *dev,phys_addr_t phys)
@@ -580,7 +580,7 @@ int dma_direct_supported(struct device *dev, u64 mask)* part of the check. */ if (IS_ENABLED (CONFIG_ZONE_DMA )) - min_mask =min_t (u64, min_mask,DMA_BIT_MASK (zone_dma_bits)); + min_mask =min_t (u64, min_mask, zone_dma_limit);return mask >=phys_to_dma_unencrypted (dev, min_mask); }
kernel/dma/pool.c +1 -1
@@ -70,7 +70,7 @@ static bool cma_in_zone(gfp_t gfp)/* CMA can't cross zone boundaries, see cma_activate_area() */ end =cma_get_base (cma) + size -1 ;if (IS_ENABLED (CONFIG_ZONE_DMA ) && (gfp &GFP_DMA )) -return end <=DMA_BIT_MASK (zone_dma_bits); +return end <= zone_dma_limit;if (IS_ENABLED (CONFIG_ZONE_DMA32 ) && (gfp &GFP_DMA32 ))return end <=DMA_BIT_MASK (32 );return true;
kernel/dma/swiotlb.c +2 -2
@@ -450,7 +450,7 @@ int swiotlb_init_late(size_t size, gfp_t gfp_mask,if (!remap) io_tlb_default_mem.can_grow = true;if (IS_ENABLED (CONFIG_ZONE_DMA ) && (gfp_mask & __GFP_DMA)) - io_tlb_default_mem.phys_limit =DMA_BIT_MASK (zone_dma_bits); + io_tlb_default_mem.phys_limit = zone_dma_limit;else if (IS_ENABLED (CONFIG_ZONE_DMA32 ) && (gfp_mask & __GFP_DMA32)) io_tlb_default_mem.phys_limit =DMA_BIT_MASK (32 );else
@@ -629,7 +629,7 @@ static struct page *swiotlb_alloc_tlb(struct device *dev, size_t bytes, } gfp &= ~GFP_ZONEMASK ; -if (phys_limit <=DMA_BIT_MASK (zone_dma_bits)) +if (phys_limit <= zone_dma_limit) gfp |= __GFP_DMA;else if (phys_limit <=DMA_BIT_MASK (32 )) gfp |= __GFP_DMA32;
Signature · 1 line
2.43.0
kernel test robot
build botRe: [PATCH v3 2/3] dma-mapping: replace zone_dma_bits by zone_dma_limit
kernel test robot noticed the following build warnings:
Hi Baruch,
kernel test robot noticed the following build warnings:
[auto build test WARNING on arm64/for-next/core]
[also build test WARNING on powerpc/next powerpc/fixes s390/features linus/master v6.11-rc1 next-20240729] [If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]
url: https://github.com/intel-lab-lkp/linux/commits/Baruch-Siach/dma-mapping-improve-DMA-zone-selection/20240729-211018 base: https://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux.git for-next/core patch link: https://lore.kernel.org/r/053fa4806a2c63efcde80caca473a8b670a2701c.1722249878.git.baruch%40tkos.co.il patch subject: [PATCH v3 2/3] dma-mapping: replace zone_dma_bits by zone_dma_limit config: arm-allnoconfig (https://download.01.org/0day-ci/archive/20240730/[email protected]/config) compiler: clang version 20.0.0git (https://github.com/llvm/llvm-project ccae7b461be339e717d02f99ac857cf0bc7d17fc) reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240730/[email protected]/reproduce)
If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/
All warnings (new ones prefixed by >>):
In file included from kernel/dma/direct.c:7:
In file included from include/linux/memblock.h:12:
In file included from include/linux/mm.h:2253:
include/linux/vmstat.h:514:36: warning: arithmetic between different enumeration types ('enum node_stat_item' and 'enum lru_list') [-Wenum-enum-conversion]
514 | return node_stat_name(NR_LRU_BASE + lru) + 3; // skip "nr_"
| ~~~~~~~~~~~ ^ ~~~
>> kernel/dma/direct.c:23:46: warning: implicit conversion from 'unsigned long long' to 'phys_addr_t' (aka 'unsigned int') changes value from 18446744073709551615 to 4294967295 [-Wconstant-conversion]
23 | phys_addr_t zone_dma_limit __ro_after_init = DMA_BIT_MASK(24);
| ~~~~~~~~~~~~~~ ^~~~~~~~~~~~~~~~
include/linux/dma-mapping.h:77:40: note: expanded from macro 'DMA_BIT_MASK'
77 | #define DMA_BIT_MASK(n) (((n) == 64) ? ~0ULL : ((1ULL<<(n))-1))
| ^~~~~2 warnings generated.
vim +23 kernel/dma/direct.c
> 7 #include <linux/memblock.h>
8 #include <linux/export.h>
9 #include <linux/mm.h>
10 #include <linux/dma-map-ops.h>
11 #include <linux/scatterlist.h>
12 #include <linux/pfn.h>
13 #include <linux/vmalloc.h>
14 #include <linux/set_memory.h>
15 #include <linux/slab.h>
16 #include "direct.h"
17
18 /*
19 * Most architectures use ZONE_DMA for the first 16 Megabytes, but some use
20 * it for entirely different regions. In that case the arch code needs to
21 * override the variable below for dma-direct to work properly.
22 */
> 23 phys_addr_t zone_dma_limit __ro_after_init = DMA_BIT_MASK(24);
24 Signature · 2 lines
0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki
Re: [PATCH v3 2/3] dma-mapping: replace zone_dma_bits by zone_dma_limit
Quoted 31 lines · kernel test robot
Hi Baruch,
kernel test robot noticed the following build warnings:
[auto build test WARNING on arm64/for-next/core]
[also build test WARNING on powerpc/next powerpc/fixes s390/features linus/master v6.11-rc1 next-20240729] [If your patch is applied to the wrong git tree, kindly drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]url: https://github.com/intel-lab-lkp/linux/commits/Baruch-Siach/dma-mapping-improve-DMA-zone-selection/20240729-211018 base: https://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux.git for-next/core patch link: https://lore.kernel.org/r/053fa4806a2c63efcde80caca473a8b670a2701c.1722249878.git.baruch%40tkos.co.il patch subject: [PATCH v3 2/3] dma-mapping: replace zone_dma_bits by zone_dma_limit config: arm-allnoconfig (https://download.01.org/0day-ci/archive/20240730/[email protected]/config) compiler: clang version 20.0.0git (https://github.com/llvm/llvm-project ccae7b461be339e717d02f99ac857cf0bc7d17fc) reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20240730/[email protected]/reproduce)If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/All warnings (new ones prefixed by >>):
In file included from
kernel/dma/direct.c:7:
In file included frominclude/linux/memblock.h:12:
In file included frominclude/linux/mm.h:2253:include/linux/vmstat.h:514:36: warning: arithmetic between different enumeration types ('enum node_stat_item' and 'enum lru_list') [-Wenum-enum-conversion] 514 | return node_stat_name(NR_LRU_BASE + lru) + 3; // skip "nr_" | ~~~~~~~~~~~ ^ ~~~
Quoted 7 lines
Quoted 1 line
Quoted 1 line
kernel/dma/direct.c:23:46: warning: implicit conversion from 'unsigned long long' to 'phys_addr_t' (aka 'unsigned int') changes value from 18446744073709551615 to 4294967295 [-Wconstant-conversion]23 | phys_addr_t zone_dma_limit __ro_after_init = DMA_BIT_MASK(24); | ~~~~~~~~~~~~~~ ^~~~~~~~~~~~~~~~ include/linux/dma-mapping.h:77:40: note: expanded from macro 'DMA_BIT_MASK' 77 | #define DMA_BIT_MASK(n) (((n) == 64) ? ~0ULL : ((1ULL<<(n))-1)) | ^~~~~2 warnings generated.
FWIW, this is likely a false positive due to an issue in Clang with the control flow graph for global variables:
https://github.com/ClangBuiltLinux/linux/issues/92
DMA_BIT_MASK() has been the biggest offender :/ If there is any way to refactor this code to avoid this, that would be great (as that has been one of our longest outstanding issues and getting it fixed in the compiler does not seem super easy at this point).
Cheers,
Nathan
Quoted 25 lines
vim +23
kernel/dma/direct.cQuoted 1 line
7 #include <
linux/memblock.h>8 #include <linux/export.h> 9 #include <linux/mm.h> 10 #include <linux/dma-map-ops.h> 11 #include <linux/scatterlist.h> 12 #include <linux/pfn.h> 13 #include <linux/vmalloc.h> 14 #include <linux/set_memory.h> 15 #include <linux/slab.h> 16 #include "direct.h" 17 18 /* 19 * Most architectures use ZONE_DMA for the first 16 Megabytes, but some use 20 * it for entirely different regions. In that case the arch code needs to 21 * override the variable below for dma-direct to work properly. 22 */Quoted 1 line
23
phys_addr_tzone_dma_limit__ro_after_init=DMA_BIT_MASK(24);24Signature · 2 lines
0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki
Re: [PATCH v3 2/3] dma-mapping: replace zone_dma_bits by zone_dma_limit
Quoted 9 lines · Nathan Chancellor
Quoted 5 lines
| ~~~~~~~~~~~~~~ ^~~~~~~~~~~~~~~~ include/linux/dma-mapping.h:77:40: note: expanded from macro 'DMA_BIT_MASK' 77 | #define DMA_BIT_MASK(n) (((n) == 64) ? ~0ULL : ((1ULL<<(n))-1)) | ^~~~~2 warnings generated.
FWIW, this is likely a false positive due to an issue in Clang with the control flow graph for global variables:
Quoted 6 lines
https://github.com/ClangBuiltLinux/linux/issues/92
DMA_BIT_MASK()has been the biggest offender :/ If there is any way to refactor this code to avoid this, that would be great (as that has been one of our longest outstanding issues and getting it fixed in the compiler does not seem super easy at this point).
I have no idea what you'd want changed here, but I'll happily take patches.
Re: [PATCH v3 2/3] dma-mapping: replace zone_dma_bits by zone_dma_limit
Quoted 12 lines · Christoph Hellwig
Quoted 11 lines · Nathan Chancellor
Quoted 5 lines
| ~~~~~~~~~~~~~~ ^~~~~~~~~~~~~~~~ include/linux/dma-mapping.h:77:40: note: expanded from macro 'DMA_BIT_MASK' 77 | #define DMA_BIT_MASK(n) (((n) == 64) ? ~0ULL : ((1ULL<<(n))-1)) | ^~~~~2 warnings generated.
FWIW, this is likely a false positive due to an issue in Clang with the control flow graph for global variables:
Quoted 7 lines
Quoted 4 lines
DMA_BIT_MASK()has been the biggest offender :/ If there is any way to refactor this code to avoid this, that would be great (as that has been one of our longest outstanding issues and getting it fixed in the compiler does not seem super easy at this point).I have no idea what you'd want changed here, but I'll happily take patches.
Unfortunately, I am not sure either... I do not see anything obviously, so perhaps it could just be avoided with the __diag() infrastructure?
kernel/dma/direct.c +5 -0
@@ -20,7 +20,12 @@* it for entirely different regions. In that case the arch code needs to * override the variable below for dma-direct to work properly. */ +__diag_push (); +__diag_ignore (clang,13 ,"-Wconstant-conversion" , +"Clang incorrectly thinks the n == 64 case in DMA_BIT_MASK() can happen here," +"which would truncate with a 32-bit phys_addr_t" );phys_addr_t zone_dma_limit__ro_after_init =DMA_BIT_MASK (24 ); +__diag_pop ();static inline dma_addr_t phys_to_dma_direct (struct device *dev,phys_addr_t phys)
Re: [PATCH v3 2/3] dma-mapping: replace zone_dma_bits by zone_dma_limit
Quoted 8 lines · Nathan Chancellor
Unfortunately, I am not sure either... I do not see anything obviously, so perhaps it could just be avoided with the
__diag()infrastructure?kernel/dma/direct.c +0
-0
kernel/dma/direct.c line 27
Quoted 8 lines
kernel/dma/direct.c +4
-0@@ -20,7 +20,12 @@* it for entirely different regions. In that case the arch code needs to * override the variable below for dma-direct to work properly. */ +__diag_push (); +__diag_ignore (clang,13 ,"-Wconstant-conversion" , +"Clang incorrectly thinks the n == 64 case in DMA_BIT_MASK() can happen here," +"which would truncate with a 32-bit phys_addr_t" );phys_addr_t zone_dma_limit__ro_after_init =DMA_BIT_MASK (24 );
So.. The code above is clearly wrong, as DMA_BIT_MASK always returns a u64, and phys_addr_t can be smaller than that. So at least in this case the warning seems perfectly valid and the code has issues because it is mixing different concepts.
Where do you see warnings like this upstream?
Re: [PATCH v3 2/3] dma-mapping: replace zone_dma_bits by zone_dma_limit
Quoted 14 lines · Christoph Hellwig
Quoted 13 lines · Nathan Chancellor
Unfortunately, I am not sure either... I do not see anything obviously, so perhaps it could just be avoided with the
__diag()infrastructure?kernel/dma/direct.c +2
-0@@ -20,7 +20,12 @@* it for entirely different regions. In that case the arch code needs to * override the variable below for dma-direct to work properly. */ +__diag_push (); +__diag_ignore(clang, 13, "-Wconstant-conversion",
Quoted 8 lines
kernel/dma/direct.c line 27
Quoted 3 lines
kernel/dma/direct.c +2
-0+"Clang incorrectly thinks the n == 64 case in DMA_BIT_MASK() can happen here," + "which would truncate with a 32-bit phys_addr_t");phys_addr_t zone_dma_limit__ro_after_init =DMA_BIT_MASK (24 );So.. The code above is clearly wrong, as
DMA_BIT_MASKalways returns a u64, andphys_addr_tcan be smaller than that. So at least in this case the warning seems perfectly valid and the code has issues because it is mixing different concepts.
Sure, that seems like a reasonable way to look at things even if the warning itself is a false positive.
Quoted 1 line
Where do you see warnings like this upstream?
I don't see this upstream, this is from patch 2 of this series:
https://lore.kernel.org/053fa4806a2c63efcde80caca473a8b670a2701c.1722249878.git.baruch@tkos.co.il/
Cheers,
Nathan
Catalin Marinas
maintainerRe: [PATCH v3 2/3] dma-mapping: replace zone_dma_bits by zone_dma_limit
In reply to Baruch SiachQuoted 10 lines · Baruch Siach
From: Catalin Marinas <[email protected]>
Hardware DMA limit might not be power of 2. When RAM range starts above 0, say 4GB, DMA limit of 30 bits should end at 5GB. A single high bit can not encode this limit.
Use direct
phys_addr_tlimit address for DMA zone limit.Signed-off-by Catalin Marinas <[email protected]>Signed-off-by Baruch Siach <[email protected]>
You should add your Co-developed-by line, the patch evolved a bit from initial my partial diff.
Quoted 16 lines
arch/arm64/mm/init.c +1
-6@@ -114,36 +114,28 @@ static void __init arch_reserve_crashkernel(void) low_size, high); } -/* -* Return the maximum physical address for a zone accessible by the given bits -* limit. If DRAM starts above 32-bit, expand the zone to the maximum -* available memory, otherwise cap it at 32-bit. -*/ -static phys_addr_t __init max_zone_phys (unsigned int zone_bits) +static phys_addr_t __init max_zone_phys (phys_addr_t zone_limit) {
arch/arm64/mm/init.c static void __init arch_reserve_crashkernel(void) · line 121
Quoted 10 lines
arch/arm64/mm/init.c +3
-7-phys_addr_t zone_mask =DMA_BIT_MASK (zone_bits); -phys_addr_t phys_start =memblock_start_of_DRAM (); - -if (phys_start >U32_MAX ) - zone_mask =PHYS_ADDR_MAX ; -else if (phys_start > zone_mask) - zone_mask =U32_MAX ; +/* We have RAM in low 32-bit area, keep DMA zone there */ +if (memblock_start_of_DRAM () <U32_MAX ) + zone_limit =min (U32_MAX , zone_limit);
Does this matter anymore? Or is it to keep ZONE_DMA below (or equal to) the ZONE_DMA32 limit?
Anyway, since this patch is about replacing zone_dma_bits with zone_dma_limit, we should not introduce functional changes. AFAICT, we need zone_limit to be set to memblock_end_of_DRAM() when phys_start is above U32_MAX. You can do the functional change in a subsequent patch once all the other refactoring has been handled.
Quoted 4 lines
arch/arm64/mm/init.c +1
-1-return min (zone_mask,memblock_end_of_DRAM () -1 ) +1 ; +return min (zone_limit,memblock_end_of_DRAM () -1 ) +1 ; }
[...]
Quoted 40 lines
include/linux/dma-direct.h +1
-1@@ -12,7 +12,7 @@#include <linux/mem_encrypt.h> #include <linux/swiotlb.h> -extern unsigned int zone_dma_bits; +extern phys_addr_t zone_dma_limit;/* * Record the mapping of CPU physical to DMA addresses for a given region. kernel/dma/direct.c +2
-2@@ -20,7 +20,7 @@* it for entirely different regions. In that case the arch code needs to * override the variable below for dma-direct to work properly. */ -unsigned int zone_dma_bits__ro_after_init =24 ; +phys_addr_t zone_dma_limit__ro_after_init =DMA_BIT_MASK (24 );static inline dma_addr_t phys_to_dma_direct (struct device *dev,phys_addr_t phys)@@ -580,7 +580,7 @@ int dma_direct_supported(struct device *dev, u64 mask)* part of the check. */ if (IS_ENABLED (CONFIG_ZONE_DMA )) - min_mask =min_t (u64, min_mask,DMA_BIT_MASK (zone_dma_bits)); + min_mask =min_t (u64, min_mask, zone_dma_limit);return mask >=phys_to_dma_unencrypted (dev, min_mask); }kernel/dma/pool.c +0
-0
kernel/dma/pool.c static bool cma_in_zone(gfp_t gfp) · line 76
Quoted 8 lines
kernel/dma/pool.c +1
-1@@ -70,7 +70,7 @@ static bool cma_in_zone(gfp_t gfp)/* CMA can't cross zone boundaries, see cma_activate_area() */ end =cma_get_base (cma) + size -1 ;if (IS_ENABLED (CONFIG_ZONE_DMA ) && (gfp &GFP_DMA )) -return end <=DMA_BIT_MASK (zone_dma_bits); +return end <= zone_dma_limit;if (IS_ENABLED (CONFIG_ZONE_DMA32 ) && (gfp &GFP_DMA32 ))return end <=DMA_BIT_MASK (32 );return true;
I haven't got to the third patch yet but with this series we can have zone_dma_limit above DMA_BIT_MASK(32). The above function could return false for GFP_DMA32 when 'end' is perfectly valid within ZONE_DMA (which implies safe for GFP_DMA32).
Quoted 14 lines
kernel/dma/swiotlb.c +1
-1@@ -450,7 +450,7 @@ int swiotlb_init_late(size_t size, gfp_t gfp_mask,if (!remap) io_tlb_default_mem.can_grow = true;if (IS_ENABLED (CONFIG_ZONE_DMA ) && (gfp_mask & __GFP_DMA)) - io_tlb_default_mem.phys_limit =DMA_BIT_MASK (zone_dma_bits); + io_tlb_default_mem.phys_limit = zone_dma_limit;else if (IS_ENABLED (CONFIG_ZONE_DMA32 ) && (gfp_mask & __GFP_DMA32)) io_tlb_default_mem.phys_limit =DMA_BIT_MASK (32 );else
kernel/dma/swiotlb.c static struct page *swiotlb_alloc_tlb(struct device *dev, size_t bytes, · line 635
Quoted 8 lines
kernel/dma/swiotlb.c +1
-1@@ -629,7 +629,7 @@ static struct page *swiotlb_alloc_tlb(struct device *dev, size_t bytes, } gfp &= ~GFP_ZONEMASK ; -if (phys_limit <=DMA_BIT_MASK (zone_dma_bits)) +if (phys_limit <= zone_dma_limit) gfp |= __GFP_DMA;else if (phys_limit <=DMA_BIT_MASK (32 )) gfp |= __GFP_DMA32;
I think this has the same issue as dma_direct_optimal_gfp_mask(). If the requested limit is strictly smaller than DMA_BIT_MASK(32) (but potentially bigger than zone_dma_limit), we should go for __GFP_DMA rather than __GFP_DMA32. You should probably fix this in the first patch as well.
As above, with this series we can end up with zone_dma_limit above DMA_BIT_MASK(32) and all these checks become confusing. Even the swiotlb_init_late() above if called with GFP_DMA32 will set a phys_limit that may not be accessible at all if the DRAM starts above 4GB. Similarly, cma_in_zone() could return false with GFP_DMA32 in similar hardware configurations.
So we either introduce a zone_dma32_limit variable and allow a 32-bit range above the start of DRAM or sanitise these sites to make sure passing GFP_DMA32 is safe - i.e. assume we only have ZONE_DMA if zone_dma_limit is above 32-bit. I prefer the latter without introducing a zone_dma32_limit.
Signature · 1 line
Catalin
PATCH v3 3/3 dma-direct: use RAM start to offset zone_dma_limit
In reply to Baruch SiachCurrent code using zone_dma_limit assume that all address range below limit is suitable for DMA. For some existing platforms this assumption is not correct. DMA range might have non zero lower limit.
Commit 791ab8b2e3db ("arm64: Ignore any DMA offsets in the max_zone_phys() calculation") made DMA/DMA32 zones span the entire RAM when RAM starts above 32-bits. This breaks hardware with DMA area that start above 32-bits. But the commit log says that "we haven't noticed any such hardware". It turns out that such hardware does exist.
One such platform has RAM starting at 32GB with an internal bus that has the following DMA limits:
#address-cells = <2>; #size-cells = <2>; dma-ranges = <0x00 0xc0000000 0x08 0x00000000 0x00 0x40000000>;
Devices under this bus can see 1GB of DMA range between 3GB-4GB in each device address space. This range is mapped to CPU memory at 32GB-33GB. With current code DMA allocations for devices under this bus are not limited to DMA area, leading to run-time allocation failure.
Add start of RAM address to zone_dma_limit to make DMA allocation for constrained devices possible.
The result is DMA zone that properly reflects the hardware constraints as follows:
[ 0.000000] Zone ranges: [ 0.000000] DMA [mem 0x0000000800000000-0x000000083fffffff] [ 0.000000] DMA32 empty [ 0.000000] Normal [mem 0x0000000840000000-0x0000000bffffffff]
Rename the dma_direct_supported() local 'min_mask' variable to better describe its use as limit.
--- kernel/dma/direct.c | 7 ++++--- kernel/dma/pool.c | 3 ++- kernel/dma/swiotlb.c | 4 ++-- 3 files changed, 8 insertions(+), 6 deletions(-)
kernel/dma/direct.c +4 -3
@@ -563,7 +563,7 @@ int dma_direct_mmap(struct device *dev, struct vm_area_struct *vma,int dma_direct_supported (struct device *dev,u64 mask) { -u64 min_mask = (max_pfn -1 ) <<PAGE_SHIFT ; +u64 min_limit = (max_pfn -1 ) <<PAGE_SHIFT ;/* * Because 32-bit DMA masks are so common we expect every architecture
@@ -580,8 +580,9 @@ int dma_direct_supported(struct device *dev, u64 mask)* part of the check. */ if (IS_ENABLED (CONFIG_ZONE_DMA )) - min_mask =min_t (u64, min_mask, zone_dma_limit); -return mask >=phys_to_dma_unencrypted (dev, min_mask); + min_limit =min_t (u64, min_limit, +memblock_start_of_DRAM () + zone_dma_limit); +return mask >=phys_to_dma_unencrypted (dev, min_limit); }/*
kernel/dma/pool.c +2 -1
@@ -12,6 +12,7 @@#include <linux/set_memory.h> #include <linux/slab.h> #include <linux/workqueue.h> +#include <linux/memblock.h> static struct gen_pool *atomic_pool_dma__ro_after_init ;static unsigned long pool_size_dma;
@@ -70,7 +71,7 @@ static bool cma_in_zone(gfp_t gfp)/* CMA can't cross zone boundaries, see cma_activate_area() */ end =cma_get_base (cma) + size -1 ;if (IS_ENABLED (CONFIG_ZONE_DMA ) && (gfp &GFP_DMA )) -return end <= zone_dma_limit; +return end <=memblock_start_of_DRAM () + zone_dma_limit;if (IS_ENABLED (CONFIG_ZONE_DMA32 ) && (gfp &GFP_DMA32 ))return end <=DMA_BIT_MASK (32 );return true;
kernel/dma/swiotlb.c +2 -2
@@ -450,7 +450,7 @@ int swiotlb_init_late(size_t size, gfp_t gfp_mask,if (!remap) io_tlb_default_mem.can_grow = true;if (IS_ENABLED (CONFIG_ZONE_DMA ) && (gfp_mask & __GFP_DMA)) - io_tlb_default_mem.phys_limit = zone_dma_limit; + io_tlb_default_mem.phys_limit =memblock_start_of_DRAM () + zone_dma_limit;else if (IS_ENABLED (CONFIG_ZONE_DMA32 ) && (gfp_mask & __GFP_DMA32)) io_tlb_default_mem.phys_limit =DMA_BIT_MASK (32 );else
@@ -629,7 +629,7 @@ static struct page *swiotlb_alloc_tlb(struct device *dev, size_t bytes, } gfp &= ~GFP_ZONEMASK ; -if (phys_limit <= zone_dma_limit) +if (phys_limit <=memblock_start_of_DRAM () + zone_dma_limit) gfp |= __GFP_DMA;else if (phys_limit <=DMA_BIT_MASK (32 )) gfp |= __GFP_DMA32;
Signature · 1 line
2.43.0
Catalin Marinas
maintainerRe: [PATCH v3 3/3] dma-direct: use RAM start to offset zone_dma_limit
Quoted 13 lines · Baruch Siach
kernel/dma/pool.c +1
-0@@ -12,6 +12,7 @@#include <linux/set_memory.h> #include <linux/slab.h> #include <linux/workqueue.h> +#include <linux/memblock.h> static struct gen_pool *atomic_pool_dma__ro_after_init ;static unsigned long pool_size_dma;
kernel/dma/pool.c static bool cma_in_zone(gfp_t gfp) · line 74
Quoted 5 lines
kernel/dma/pool.c +1
-1@@ -70,7 +71,7 @@ static bool cma_in_zone(gfp_t gfp)/* CMA can't cross zone boundaries, see cma_activate_area() */ end =cma_get_base (cma) + size -1 ;if (IS_ENABLED (CONFIG_ZONE_DMA ) && (gfp &GFP_DMA )) -return end <= zone_dma_limit; +return end <=memblock_start_of_DRAM () + zone_dma_limit;
I think this patch is entirely wrong. After the previous patch, zone_dma_limit is already a physical/CPU address, not some offset or range - of_dma_get_max_cpu_address() returns the absolute physical address. Adding memblock_start_of_DRAM() to it does not make any sense. It made sense when we had zone_dma_bits but since we are trying to move away from bitmasks to absolute CPU addresses, zone_dma_limit already includes the start of DRAM.
What problems do you see without this patch? Is it because DMA_BIT_MASK(32) can be lower than zone_dma_limit as I mentioned on the previous patch?
Signature · 1 line
Catalin