Skip to contentSkip to conversation

iommu/io-pgtable-arm-v7s: Use DMA32 zone for page tables

PATCH In review Nicolas Boichat 6 participants 17 messages Dec 10, 2018 – Mar 20, 2019 1 message not in the archive

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

v6 cover iommu/io-pgtable-arm-v7s: Use DMA32 zone for page tables

This is a follow-up to the discussion in [1], [2].

IOMMUs using ARMv7 short-descriptor format require page tables (level 1 and 2) to be allocated within the first 4GB of RAM, even on 64-bit systems.

For L1 tables that are bigger than a page, we can just use __get_free_pages with GFP_DMA32 (on arm64 systems only, arm would still use GFP_DMA).

For L2 tables that only take 1KB, it would be a waste to allocate a full page, so we considered 3 approaches:
1. This series, adding support for GFP_DMA32 slab caches.
2. genalloc, which requires pre-allocating the maximum number of L2 page

    tables (4096, so 4MB of memory).

3. page_frag, which is not very memory-efficient as it is unable to reuse freed fragments until the whole page is freed. [3]

This series is the most memory-efficient approach.

stable@ note:
We confirmed that this is a regression, and IOMMU errors happen on 4.19 and linux-next/master on MT8173 (elm, Acer Chromebook R13). The issue most likely starts from commit ad67f5a6545f ("arm64: replace ZONE_DMA with ZONE_DMA32"), i.e. 4.15, and presumably breaks a number of Mediatek platforms (and maybe others?).

[1] https://lists.linuxfoundation.org/pipermail/iommu/2018-November/030876.html
[2] https://lists.linuxfoundation.org/pipermail/iommu/2018-December/031696.html
[3] https://patchwork.codeaurora.org/patch/671639/

Changes since v1:
- Add support for SLAB_CACHE_DMA32 in slab and slub (patches 1/2)
- iommu/io-pgtable-arm-v7s (patch 3):
- Changed approach to use SLAB_CACHE_DMA32 added by the previous commit.
- Use DMA or DMA32 depending on the architecture (DMA for arm,

     DMA32 for arm64).

Changes since v2:
- Reworded and expanded commit messages
- Added cache_dma32 documentation in PATCH 2/3.

v3 used the page_frag approach, see [3].

Changes since v4:
- Dropped change that removed GFP_DMA32 from GFP_SLAB_BUG_MASK:
instead we can just call kmem_cache_*alloc without GFP_DMA32 parameter. This also means that we can drop PATCH v4 1/3, as we do not make any changes in GFP flag verification.
- Dropped hunks that added cache_dma32 sysfs file, and moved the hunks to PATCH v5 3/3, so that maintainer can decide whether to pick the change independently.

Changes since v5:
- Rename ARM_V7S_TABLE_SLAB_CACHE to ARM_V7S_TABLE_SLAB_FLAGS.
- Add stable@ to cc.

Nicolas Boichat (3):
mm: Add support for kmem caches in DMA32 zone iommu/io-pgtable-arm-v7s: Request DMA32 memory, and improve debugging
mm: Add /sys/kernel/slab/cache/cache_dma32

 Documentation/ABI/testing/sysfs-kernel-slab |  9 +++++++++
 drivers/iommu/io-pgtable-arm-v7s.c          | 19 +++++++++++++++----
 include/linux/slab.h                        |  2 ++
 mm/slab.c                                   |  2 ++
 mm/slab.h                                   |  3 ++-
 mm/slab_common.c                            |  2 +-
 mm/slub.c                                   | 16 ++++++++++++++++
 tools/vm/slabinfo.c                         |  7 ++++++-
 8 files changed, 53 insertions(+), 7 deletions(-)
Signature · 1 line
2.20.0.rc2.403.gdbc3b29805-goog
#1 lore raw

PATCH v6 1/3 mm: Add support for kmem caches in DMA32 zone

IOMMUs using ARMv7 short-descriptor format require page tables to be allocated within the first 4GB of RAM, even on 64-bit systems. On arm64, this is done by passing GFP_DMA32 flag to memory allocation functions.

For IOMMU L2 tables that only take 1KB, it would be a waste to allocate a full page using get_free_pages, so we considered 3 approaches:
1. This patch, adding support for GFP_DMA32 slab caches.
2. genalloc, which requires pre-allocating the maximum number of L2 page tables (4096, so 4MB of memory).
3. page_frag, which is not very memory-efficient as it is unable to reuse freed fragments until the whole page is freed.

This change makes it possible to create a custom cache in DMA32 zone using kmem_cache_create, then allocate memory using kmem_cache_alloc.

We do not create a DMA32 kmalloc cache array, as there are currently no users of kmalloc(..., GFP_DMA32). These calls will continue to trigger a warning, as we keep GFP_DMA32 in GFP_SLAB_BUG_MASK.

This implies that calls to kmem_cache_*alloc on a SLAB_CACHE_DMA32 kmem_cache must _not_ use GFP_DMA32 (it is anyway redundant and unnecessary).

---

Changes since v2:
- Clarified commit message
- Add entry in sysfs-kernel-slab to document the new sysfs file

(v3 used the page_frag approach)

Changes since v4:
- Added details to commit message
- Dropped change that removed GFP_DMA32 from GFP_SLAB_BUG_MASK:
instead we can just call kmem_cache_*alloc without GFP_DMA32 parameter. This also means that we can drop PATCH 1/3, as we do not make any changes in GFP flag verification.
- Dropped hunks that added cache_dma32 sysfs file, and moved the hunks to PATCH 3/3, so that maintainer can decide whether to pick the change independently.

(no change since v5)

 include/linux/slab.h | 2 ++
 mm/slab.c            | 2 ++
 mm/slab.h            | 3 ++-
 mm/slab_common.c     | 2 +-
 mm/slub.c            | 5 +++++
 5 files changed, 12 insertions(+), 2 deletions(-)
include/linux/slab.h +2 -0
index 11b45f7ae4057c..9449b19c5f107a 100644
@@ -32,6 +32,8 @@
 #define SLAB_HWCACHE_ALIGN	((slab_flags_t __force)0x00002000U)
 /* Use GFP_DMA memory */
 #define SLAB_CACHE_DMA		((slab_flags_t __force)0x00004000U)
+/* Use GFP_DMA32 memory */
+#define SLAB_CACHE_DMA32	((slab_flags_t __force)0x00008000U)
 /* DEBUG: Store the last owner for bug hunting */
 #define SLAB_STORE_USER		((slab_flags_t __force)0x00010000U)
 /* Panic if kmem_cache_create() fails */
mm/slab.c +2 -0
index 73fe23e649c91a..124f8c556d27fb 100644
@@ -2109,6 +2109,8 @@ int __kmem_cache_create(struct kmem_cache *cachep, slab_flags_t flags)
 	cachep->allocflags = __GFP_COMP;
 	if (flags & SLAB_CACHE_DMA)
 		cachep->allocflags |= GFP_DMA;
+	if (flags & SLAB_CACHE_DMA32)
+		cachep->allocflags |= GFP_DMA32;
 	if (flags & SLAB_RECLAIM_ACCOUNT)
 		cachep->allocflags |= __GFP_RECLAIMABLE;
 	cachep->size = size;
mm/slab.h +2 -1
index 4190c24ef0e9df..fcf717e12f0a86 100644
@@ -127,7 +127,8 @@ static inline slab_flags_t kmem_cache_flags(unsigned int object_size,
 
 
 /* Legal flag mask for kmem_cache_create(), for various configurations */
-#define SLAB_CORE_FLAGS (SLAB_HWCACHE_ALIGN | SLAB_CACHE_DMA | SLAB_PANIC | \
+#define SLAB_CORE_FLAGS (SLAB_HWCACHE_ALIGN | SLAB_CACHE_DMA | \
+			 SLAB_CACHE_DMA32 | SLAB_PANIC | \
 			 SLAB_TYPESAFE_BY_RCU | SLAB_DEBUG_OBJECTS )
 
 #if defined(CONFIG_DEBUG_SLAB)
mm/slab_common.c +1 -1
index 70b0cc85db67f8..18b7b809c8d064 100644
@@ -53,7 +53,7 @@ static DECLARE_WORK(slab_caches_to_rcu_destroy_work,
 		SLAB_FAILSLAB | SLAB_KASAN)
 
 #define SLAB_MERGE_SAME (SLAB_RECLAIM_ACCOUNT | SLAB_CACHE_DMA | \
-			 SLAB_ACCOUNT)
+			 SLAB_CACHE_DMA32 | SLAB_ACCOUNT)
 
 /*
  * Merge control. If this is set then no merging of slab caches will occur.
mm/slub.c +5 -0
index c229a9b7dd5448..4caadb926838ef 100644
@@ -3583,6 +3583,9 @@ static int calculate_sizes(struct kmem_cache *s, int forced_order)
 	if (s->flags & SLAB_CACHE_DMA)
 		s->allocflags |= GFP_DMA;
 
+	if (s->flags & SLAB_CACHE_DMA32)
+		s->allocflags |= GFP_DMA32;
+
 	if (s->flags & SLAB_RECLAIM_ACCOUNT)
 		s->allocflags |= __GFP_RECLAIMABLE;
 
@@ -5671,6 +5674,8 @@ static char *create_unique_id(struct kmem_cache *s)
 	 */
 	if (s->flags & SLAB_CACHE_DMA)
 		*p++ = 'd';
+	if (s->flags & SLAB_CACHE_DMA32)
+		*p++ = 'D';
 	if (s->flags & SLAB_RECLAIM_ACCOUNT)
 		*p++ = 'a';
 	if (s->flags & SLAB_CONSISTENCY_CHECKS)
Signature · 1 line
2.20.0.rc2.403.gdbc3b29805-goog

A message that is not in the archive (sent off-list or not yet received): [email protected]

#3 lore raw

Re: [PATCH v6 1/3] mm: Add support for kmem caches in DMA32 zone

Hi Sasha,

Quoted 65 lines · Sasha Levin

Hi,

[This is an automated email]

This commit has been processed because it contains a -stable tag. The stable tag indicates that it's relevant for the following trees: all

The bot has tested the following trees: v4.19.8, v4.14.87, v4.9.144, v4.4.166, v3.18.128,

v4.19.8: Build OK!
v4.14.87: Failed to apply! Possible dependencies:

    4fd0b46e8987 ("slab, slub, slob: convert slab_flags_t to 32-bit")
    a3ba07444782 ("mm/slab.c: only set __GFP_RECLAIMABLE once")
    d50112edde1d ("slab, slub, slob: add slab_flags_t")

v4.9.144: Failed to apply! Possible dependencies:

    04d348ae3f0a ("drm/i915/gvt: vGPU display virtualization")
    12d14cc43b34 ("drm/i915/gvt: Introduce a framework for tracking HW registers.")
    1b36595ffb35 ("drm/i915: Show RING registers through debugfs")
    28a60dee2ce6 ("drm/i915/gvt: vGPU HW resource management")
    3f728236c516 ("drm/i915/gvt: trace stub")
    4fd0b46e8987 ("slab, slub, slob: convert slab_flags_t to 32-bit")
    579cea5f30f2 ("drm/i915/gvt: golden virtual HW state management")
    5f0d5a3ae7cf ("mm: Rename SLAB_DESTROY_BY_RCU to SLAB_TYPESAFE_BY_RCU")
    73cb97010d4f ("drm/i915: Combine seqno + tracking into a global timeline struct")
    82d375d1b568 ("drm/i915/gvt: Introduce basic vGPU life cycle management")
    8453d674ae7e ("drm/i915/gvt: vGPU execlist virtualization")
    a933568eb61d ("drm/i915: Tidy slab cache allocations")
    c8fe6a6811a7 ("drm/i915/gvt: vGPU interrupt virtualization.")
    d50112edde1d ("slab, slub, slob: add slab_flags_t")
    e39c5add3221 ("drm/i915/gvt: vGPU MMIO virtualization")
    e473405783c0 ("drm/i915/gvt: vGPU workload scheduler")
    e95433c73a11 ("drm/i915: Rearrange i915_wait_request() accounting with callers")

v4.4.166: Failed to apply! Possible dependencies:
10b2e9e8e808 ("mm/slab: factor out debugging initialization in cache_init_objs()") 40b44137971c ("mm/slab: clean up DEBUG_PAGEALLOC processing code") 4fd0b46e8987 ("slab, slub, slob: convert slab_flags_t to 32-bit") 505f5dcb1c41 ("mm, kasan: add GFP flags to KASAN API")
55834c59098d ("mm: kasan: initial memory quarantine implementation") 5d097056c9a0 ("kmemcg: account certain kmem allocations to memcg") 7ed2f9e66385 ("mm, kasan: SLAB support")
832a15d209cd ("mm/slab: align cache size first before determination of OFF_SLAB candidate") a307ebd468e0 ("mm/slab: activate debug_pagealloc in SLAB when it is actually enabled") d50112edde1d ("slab, slub, slob: add slab_flags_t")

v3.18.128: Failed to apply! Possible dependencies:
081fa9dffcde ("staging/lustre/llite: remove llite proc root on init failure") 12cf89b550d1 ("livepatch: rename config to CONFIG_LIVEPATCH")
13d1cf7e7025 ("livepatch: samples: add sample live patching module")
31e7213ab21e ("drivers: staging: lustre: Fix "spaces required around that '='" errors") 32b7eb877165 ("livepatch: change ARCH_HAVE_LIVE_PATCHING to HAVE_LIVE_PATCHING") 4fd0b46e8987 ("slab, slub, slob: convert slab_flags_t to 32-bit")
5aaeb5c01c5b ("x86/fpu, sched: Introduce CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT and use it on x86") 5d097056c9a0 ("kmemcg: account certain kmem allocations to memcg")
6471b825c41e ("x86/kconfig: Reorganize arch feature Kconfig select's") 6e0a0ea12962 ("ACPI / sleep: Introduce CONFIG_ACPI_SYSTEM_POWER_STATES_SUPPORT") 83ac237a950e ("livepatch: kconfig: use bool instead of boolean")
957e3facd147 ("gcov: enable GCOV_PROFILE_ALL from ARCH Kconfigs")
b700e7f03df5 ("livepatch: kernel: add support for live patching")
b9dfe0bed999 ("livepatch: handle ancient compilers with more grace")
d50112edde1d ("slab, slub, slob: add slab_flags_t")

Quoted 1 line

How should we proceed with this patch?

I'm not sure how we need to handle this.

2/3 in this series (https://patchwork.kernel.org/patch/10720495/) is the patch that matters, and has correct Fixes/Cc tag (this only needs to apply >=4.15).

I added Cc: stable@ to this patch, as it is a dependency of 2/3. Should I remove Cc: stable from this patch? Or add Fixes: line?

Thanks,

Nicolas

Quoted 3 lines

--
Thanks,
Sasha

maintainer
#4 lore raw

Re: [PATCH v6 1/3] mm: Add support for kmem caches in DMA32 zone

Quoted 70 lines · Nicolas Boichat

Hi Sasha,

Quoted 65 lines · Sasha Levin

Hi,

[This is an automated email]

This commit has been processed because it contains a -stable tag. The stable tag indicates that it's relevant for the following trees: all

The bot has tested the following trees: v4.19.8, v4.14.87, v4.9.144, v4.4.166, v3.18.128,

v4.19.8: Build OK!
v4.14.87: Failed to apply! Possible dependencies:

    4fd0b46e8987 ("slab, slub, slob: convert slab_flags_t to 32-bit")
    a3ba07444782 ("mm/slab.c: only set __GFP_RECLAIMABLE once")
    d50112edde1d ("slab, slub, slob: add slab_flags_t")

v4.9.144: Failed to apply! Possible dependencies:

    04d348ae3f0a ("drm/i915/gvt: vGPU display virtualization")
    12d14cc43b34 ("drm/i915/gvt: Introduce a framework for tracking HW registers.")
    1b36595ffb35 ("drm/i915: Show RING registers through debugfs")
    28a60dee2ce6 ("drm/i915/gvt: vGPU HW resource management")
    3f728236c516 ("drm/i915/gvt: trace stub")
    4fd0b46e8987 ("slab, slub, slob: convert slab_flags_t to 32-bit")
    579cea5f30f2 ("drm/i915/gvt: golden virtual HW state management")
    5f0d5a3ae7cf ("mm: Rename SLAB_DESTROY_BY_RCU to SLAB_TYPESAFE_BY_RCU")
    73cb97010d4f ("drm/i915: Combine seqno + tracking into a global timeline struct")
    82d375d1b568 ("drm/i915/gvt: Introduce basic vGPU life cycle management")
    8453d674ae7e ("drm/i915/gvt: vGPU execlist virtualization")
    a933568eb61d ("drm/i915: Tidy slab cache allocations")
    c8fe6a6811a7 ("drm/i915/gvt: vGPU interrupt virtualization.")
    d50112edde1d ("slab, slub, slob: add slab_flags_t")
    e39c5add3221 ("drm/i915/gvt: vGPU MMIO virtualization")
    e473405783c0 ("drm/i915/gvt: vGPU workload scheduler")
    e95433c73a11 ("drm/i915: Rearrange i915_wait_request() accounting with callers")

v4.4.166: Failed to apply! Possible dependencies:
10b2e9e8e808 ("mm/slab: factor out debugging initialization in cache_init_objs()") 40b44137971c ("mm/slab: clean up DEBUG_PAGEALLOC processing code") 4fd0b46e8987 ("slab, slub, slob: convert slab_flags_t to 32-bit") 505f5dcb1c41 ("mm, kasan: add GFP flags to KASAN API")
55834c59098d ("mm: kasan: initial memory quarantine implementation") 5d097056c9a0 ("kmemcg: account certain kmem allocations to memcg") 7ed2f9e66385 ("mm, kasan: SLAB support")
832a15d209cd ("mm/slab: align cache size first before determination of OFF_SLAB candidate") a307ebd468e0 ("mm/slab: activate debug_pagealloc in SLAB when it is actually enabled") d50112edde1d ("slab, slub, slob: add slab_flags_t")

v3.18.128: Failed to apply! Possible dependencies:
081fa9dffcde ("staging/lustre/llite: remove llite proc root on init failure") 12cf89b550d1 ("livepatch: rename config to CONFIG_LIVEPATCH")
13d1cf7e7025 ("livepatch: samples: add sample live patching module")
31e7213ab21e ("drivers: staging: lustre: Fix "spaces required around that '='" errors") 32b7eb877165 ("livepatch: change ARCH_HAVE_LIVE_PATCHING to HAVE_LIVE_PATCHING") 4fd0b46e8987 ("slab, slub, slob: convert slab_flags_t to 32-bit")
5aaeb5c01c5b ("x86/fpu, sched: Introduce CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT and use it on x86") 5d097056c9a0 ("kmemcg: account certain kmem allocations to memcg")
6471b825c41e ("x86/kconfig: Reorganize arch feature Kconfig select's") 6e0a0ea12962 ("ACPI / sleep: Introduce CONFIG_ACPI_SYSTEM_POWER_STATES_SUPPORT") 83ac237a950e ("livepatch: kconfig: use bool instead of boolean")
957e3facd147 ("gcov: enable GCOV_PROFILE_ALL from ARCH Kconfigs")
b700e7f03df5 ("livepatch: kernel: add support for live patching")
b9dfe0bed999 ("livepatch: handle ancient compilers with more grace")
d50112edde1d ("slab, slub, slob: add slab_flags_t")

Quoted 1 line

How should we proceed with this patch?

Quoted 8 lines

I'm not sure how we need to handle this.

2/3 in this series (https://patchwork.kernel.org/patch/10720495/) is the patch that matters, and has correct Fixes/Cc tag (this only needs to apply >=4.15).

I added Cc: stable@ to this patch, as it is a dependency of 2/3. Should I remove Cc: stable from this patch? Or add Fixes: line?

It's fine if you want to keep it. This mail is just a heads-up saying that it won't apply on <=4.14 kernels without a backport.

--
Thanks,
Sasha

#5 lore raw

Re: [PATCH v6 1/3] mm: Add support for kmem caches in DMA32 zone

Quoted 78 lines · Sasha Levin
Quoted 76 lines · Nicolas Boichat

Hi Sasha,

Quoted 65 lines · Sasha Levin

Hi,

[This is an automated email]

This commit has been processed because it contains a -stable tag. The stable tag indicates that it's relevant for the following trees: all

The bot has tested the following trees: v4.19.8, v4.14.87, v4.9.144, v4.4.166, v3.18.128,

v4.19.8: Build OK!
v4.14.87: Failed to apply! Possible dependencies:

    4fd0b46e8987 ("slab, slub, slob: convert slab_flags_t to 32-bit")
    a3ba07444782 ("mm/slab.c: only set __GFP_RECLAIMABLE once")
    d50112edde1d ("slab, slub, slob: add slab_flags_t")

v4.9.144: Failed to apply! Possible dependencies:

    04d348ae3f0a ("drm/i915/gvt: vGPU display virtualization")
    12d14cc43b34 ("drm/i915/gvt: Introduce a framework for tracking HW registers.")
    1b36595ffb35 ("drm/i915: Show RING registers through debugfs")
    28a60dee2ce6 ("drm/i915/gvt: vGPU HW resource management")
    3f728236c516 ("drm/i915/gvt: trace stub")
    4fd0b46e8987 ("slab, slub, slob: convert slab_flags_t to 32-bit")
    579cea5f30f2 ("drm/i915/gvt: golden virtual HW state management")
    5f0d5a3ae7cf ("mm: Rename SLAB_DESTROY_BY_RCU to SLAB_TYPESAFE_BY_RCU")
    73cb97010d4f ("drm/i915: Combine seqno + tracking into a global timeline struct")
    82d375d1b568 ("drm/i915/gvt: Introduce basic vGPU life cycle management")
    8453d674ae7e ("drm/i915/gvt: vGPU execlist virtualization")
    a933568eb61d ("drm/i915: Tidy slab cache allocations")
    c8fe6a6811a7 ("drm/i915/gvt: vGPU interrupt virtualization.")
    d50112edde1d ("slab, slub, slob: add slab_flags_t")
    e39c5add3221 ("drm/i915/gvt: vGPU MMIO virtualization")
    e473405783c0 ("drm/i915/gvt: vGPU workload scheduler")
    e95433c73a11 ("drm/i915: Rearrange i915_wait_request() accounting with callers")

v4.4.166: Failed to apply! Possible dependencies:
10b2e9e8e808 ("mm/slab: factor out debugging initialization in cache_init_objs()") 40b44137971c ("mm/slab: clean up DEBUG_PAGEALLOC processing code") 4fd0b46e8987 ("slab, slub, slob: convert slab_flags_t to 32-bit") 505f5dcb1c41 ("mm, kasan: add GFP flags to KASAN API")
55834c59098d ("mm: kasan: initial memory quarantine implementation") 5d097056c9a0 ("kmemcg: account certain kmem allocations to memcg") 7ed2f9e66385 ("mm, kasan: SLAB support")
832a15d209cd ("mm/slab: align cache size first before determination of OFF_SLAB candidate") a307ebd468e0 ("mm/slab: activate debug_pagealloc in SLAB when it is actually enabled") d50112edde1d ("slab, slub, slob: add slab_flags_t")

v3.18.128: Failed to apply! Possible dependencies:
081fa9dffcde ("staging/lustre/llite: remove llite proc root on init failure") 12cf89b550d1 ("livepatch: rename config to CONFIG_LIVEPATCH")
13d1cf7e7025 ("livepatch: samples: add sample live patching module")
31e7213ab21e ("drivers: staging: lustre: Fix "spaces required around that '='" errors") 32b7eb877165 ("livepatch: change ARCH_HAVE_LIVE_PATCHING to HAVE_LIVE_PATCHING") 4fd0b46e8987 ("slab, slub, slob: convert slab_flags_t to 32-bit")
5aaeb5c01c5b ("x86/fpu, sched: Introduce CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT and use it on x86") 5d097056c9a0 ("kmemcg: account certain kmem allocations to memcg")
6471b825c41e ("x86/kconfig: Reorganize arch feature Kconfig select's") 6e0a0ea12962 ("ACPI / sleep: Introduce CONFIG_ACPI_SYSTEM_POWER_STATES_SUPPORT") 83ac237a950e ("livepatch: kconfig: use bool instead of boolean")
957e3facd147 ("gcov: enable GCOV_PROFILE_ALL from ARCH Kconfigs")
b700e7f03df5 ("livepatch: kernel: add support for live patching")
b9dfe0bed999 ("livepatch: handle ancient compilers with more grace")
d50112edde1d ("slab, slub, slob: add slab_flags_t")

Quoted 1 line

How should we proceed with this patch?

I'm not sure how we need to handle this.

2/3 in this series (https://patchwork.kernel.org/patch/10720495/) is the patch that matters, and has correct Fixes/Cc tag (this only needs to apply >=4.15).

Quoted 5 lines
Quoted 2 lines

I added Cc: stable@ to this patch, as it is a dependency of 2/3. Should I remove Cc: stable from this patch? Or add Fixes: line?

It's fine if you want to keep it. This mail is just a heads-up saying that it won't apply on <=4.14 kernels without a backport.

Noted. Thanks!

Quoted 3 lines

--
Thanks,
Sasha

#6 lore raw

PATCH v6 2/3 iommu/io-pgtable-arm-v7s: Request DMA32 memory, and improve debugging

In reply to Nicolas Boichat

IOMMUs using ARMv7 short-descriptor format require page tables (level 1 and 2) to be allocated within the first 4GB of RAM, even on 64-bit systems.

For level 1/2 pages, ensure GFP_DMA32 is used if CONFIG_ZONE_DMA32 is defined (e.g. on arm64 platforms).

For level 2 pages, allocate a slab cache in SLAB_CACHE_DMA32. Note that we do not explicitly pass GFP_DMA[32] to kmem_cache_zalloc, as this is not strictly necessary, and would cause a warning in mm/sl*b.c, as we did not update GFP_SLAB_BUG_MASK.

Also, print an error when the physical address does not fit in 32-bit, to make debugging easier in the future.

Fixes ad67f5a6545f ("arm64: replace ZONE_DMA with ZONE_DMA32")

---

Changes since v2:
- Commit message

(v3 used the page_frag approach)

Changes since v4:
- Do not pass ARM_V7S_TABLE_GFP_DMA to kmem_cache_zalloc, as this is unnecessary, and would trigger a warning.

Changes since v5:
- Rename ARM_V7S_TABLE_SLAB_CACHE to ARM_V7S_TABLE_SLAB_FLAGS.

 drivers/iommu/io-pgtable-arm-v7s.c | 19 +++++++++++++++----
 1 file changed, 15 insertions(+), 4 deletions(-)
drivers/iommu/io-pgtable-arm-v7s.c +15 -4
index 445c3bde04800c..d2fdb192f7610f 100644
@@ -161,6 +161,14 @@
 
 #define ARM_V7S_TCR_PD1			BIT(5)
 
+#ifdef CONFIG_ZONE_DMA32
+#define ARM_V7S_TABLE_GFP_DMA GFP_DMA32
+#define ARM_V7S_TABLE_SLAB_FLAGS SLAB_CACHE_DMA32
+#else
+#define ARM_V7S_TABLE_GFP_DMA GFP_DMA
+#define ARM_V7S_TABLE_SLAB_FLAGS SLAB_CACHE_DMA
+#endif
+
 typedef u32 arm_v7s_iopte;
 
 static bool selftest_running;
@@ -198,13 +206,16 @@ static void *__arm_v7s_alloc_table(int lvl, gfp_t gfp,
 	void *table = NULL;
 
 	if (lvl == 1)
-		table = (void *)__get_dma_pages(__GFP_ZERO, get_order(size));
+		table = (void *)__get_free_pages(
+			__GFP_ZERO | ARM_V7S_TABLE_GFP_DMA, get_order(size));
 	else if (lvl == 2)
-		table = kmem_cache_zalloc(data->l2_tables, gfp | GFP_DMA);
+		table = kmem_cache_zalloc(data->l2_tables, gfp);
 	phys = virt_to_phys(table);
-	if (phys != (arm_v7s_iopte)phys)
+	if (phys != (arm_v7s_iopte)phys) {
 		/* Doesn't fit in PTE */
+		dev_err(dev, "Page table does not fit in PTE: %pa", &phys);
 		goto out_free;
+	}
 	if (table && !(cfg->quirks & IO_PGTABLE_QUIRK_NO_DMA)) {
 		dma = dma_map_single(dev, table, size, DMA_TO_DEVICE);
 		if (dma_mapping_error(dev, dma))
@@ -737,7 +748,7 @@ static struct io_pgtable *arm_v7s_alloc_pgtable(struct io_pgtable_cfg *cfg,
 	data->l2_tables = kmem_cache_create("io-pgtable_armv7s_l2",
 					    ARM_V7S_TABLE_SIZE(2),
 					    ARM_V7S_TABLE_SIZE(2),
-					    SLAB_CACHE_DMA, NULL);
+					    ARM_V7S_TABLE_SLAB_FLAGS, NULL);
 	if (!data->l2_tables)
 		goto out_free_data;
 
Signature · 1 line
2.20.0.rc2.403.gdbc3b29805-goog
Acked-by on v6 2/3
#7 lore raw

Re: [PATCH v6 2/3] iommu/io-pgtable-arm-v7s: Request DMA32 memory, and improve debugging

Quoted 12 lines · Nicolas Boichat

IOMMUs using ARMv7 short-descriptor format require page tables (level 1 and 2) to be allocated within the first 4GB of RAM, even on 64-bit systems.

For level 1/2 pages, ensure GFP_DMA32 is used if CONFIG_ZONE_DMA32 is defined (e.g. on arm64 platforms).

For level 2 pages, allocate a slab cache in SLAB_CACHE_DMA32. Note that we do not explicitly pass GFP_DMA[32] to kmem_cache_zalloc, as this is not strictly necessary, and would cause a warning in mm/sl*b.c, as we did not update GFP_SLAB_BUG_MASK.

Quoted 7 lines

Also, print an error when the physical address does not fit in 32-bit, to make debugging easier in the future.

Fixes ad67f5a6545f ("arm64: replace ZONE_DMA with ZONE_DMA32")

---

Will

Acked-by on v6 2/3
#8 lore raw

Re: [PATCH v6 2/3] iommu/io-pgtable-arm-v7s: Request DMA32 memory, and improve debugging

In reply to Nicolas Boichat
Quoted 12 lines · Nicolas Boichat

IOMMUs using ARMv7 short-descriptor format require page tables (level 1 and 2) to be allocated within the first 4GB of RAM, even on 64-bit systems.

For level 1/2 pages, ensure GFP_DMA32 is used if CONFIG_ZONE_DMA32 is defined (e.g. on arm64 platforms).

For level 2 pages, allocate a slab cache in SLAB_CACHE_DMA32. Note that we do not explicitly pass GFP_DMA[32] to kmem_cache_zalloc, as this is not strictly necessary, and would cause a warning in mm/sl*b.c, as we did not update GFP_SLAB_BUG_MASK.

Quoted 6 lines

Also, print an error when the physical address does not fit in 32-bit, to make debugging easier in the future.

Fixes ad67f5a6545f ("arm64: replace ZONE_DMA with ZONE_DMA32")

Assuming you're routing all of this via akpm:

Will

#9 lore raw

PATCH v6 3/3 mm: Add /sys/kernel/slab/cache/cache_dma32

In reply to Nicolas Boichat

A previous patch in this series adds support for SLAB_CACHE_DMA32 kmem caches. This adds the corresponding /sys/kernel/slab/cache/cache_dma32 entries, and fixes slabinfo tool.

---

There were different opinions on whether this sysfs entry should be added, so I'll leave it up to the mm/slub maintainers to decide whether they want to pick this up, or drop it.

No change since v5.

 Documentation/ABI/testing/sysfs-kernel-slab |  9 +++++++++
 mm/slub.c                                   | 11 +++++++++++
 tools/vm/slabinfo.c                         |  7 ++++++-
 3 files changed, 26 insertions(+), 1 deletion(-)
Documentation/ABI/testing/sysfs-kernel-slab +9 -0
index 29601d93a1c2ea..d742c6cfdffbe9 100644
@@ -106,6 +106,15 @@ Description:
 		are from ZONE_DMA.
 		Available when CONFIG_ZONE_DMA is enabled.
 
+What:		/sys/kernel/slab/cache/cache_dma32
+Date:		December 2018
+KernelVersion:	4.21
+Contact:	Nicolas Boichat <[email protected]>
+Description:
+		The cache_dma32 file is read-only and specifies whether objects
+		are from ZONE_DMA32.
+		Available when CONFIG_ZONE_DMA32 is enabled.
+
 What:		/sys/kernel/slab/cache/cpu_slabs
 Date:		May 2007
 KernelVersion:	2.6.22
mm/slub.c +11 -0
index 4caadb926838ef..840f3719d9d543 100644
@@ -5104,6 +5104,14 @@ static ssize_t cache_dma_show(struct kmem_cache *s, char *buf)
 SLAB_ATTR_RO(cache_dma);
 #endif
 
+#ifdef CONFIG_ZONE_DMA32
+static ssize_t cache_dma32_show(struct kmem_cache *s, char *buf)
+{
+	return sprintf(buf, "%d\n", !!(s->flags & SLAB_CACHE_DMA32));
+}
+SLAB_ATTR_RO(cache_dma32);
+#endif
+
 static ssize_t usersize_show(struct kmem_cache *s, char *buf)
 {
 	return sprintf(buf, "%u\n", s->usersize);
@@ -5444,6 +5452,9 @@ static struct attribute *slab_attrs[] = {
 #ifdef CONFIG_ZONE_DMA
 	&cache_dma_attr.attr,
 #endif
+#ifdef CONFIG_ZONE_DMA32
+	&cache_dma32_attr.attr,
+#endif
 #ifdef CONFIG_NUMA
 	&remote_node_defrag_ratio_attr.attr,
 #endif
tools/vm/slabinfo.c +6 -1
index 334b16db0ebbe9..4ee1bf6e498dfa 100644
@@ -29,7 +29,7 @@ struct slabinfo {
 	char *name;
 	int alias;
 	int refs;
-	int aliases, align, cache_dma, cpu_slabs, destroy_by_rcu;
+	int aliases, align, cache_dma, cache_dma32, cpu_slabs, destroy_by_rcu;
 	unsigned int hwcache_align, object_size, objs_per_slab;
 	unsigned int sanity_checks, slab_size, store_user, trace;
 	int order, poison, reclaim_account, red_zone;
@@ -531,6 +531,8 @@ static void report(struct slabinfo *s)
 		printf("** Hardware cacheline aligned\n");
 	if (s->cache_dma)
 		printf("** Memory is allocated in a special DMA zone\n");
+	if (s->cache_dma32)
+		printf("** Memory is allocated in a special DMA32 zone\n");
 	if (s->destroy_by_rcu)
 		printf("** Slabs are destroyed via RCU\n");
 	if (s->reclaim_account)
@@ -599,6 +601,8 @@ static void slabcache(struct slabinfo *s)
 		*p++ = '*';
 	if (s->cache_dma)
 		*p++ = 'd';
+	if (s->cache_dma32)
+		*p++ = 'D';
 	if (s->hwcache_align)
 		*p++ = 'A';
 	if (s->poison)
@@ -1205,6 +1209,7 @@ static void read_slab_dir(void)
 			slab->aliases = get_obj("aliases");
 			slab->align = get_obj("align");
 			slab->cache_dma = get_obj("cache_dma");
+			slab->cache_dma32 = get_obj("cache_dma32");
 			slab->cpu_slabs = get_obj("cpu_slabs");
 			slab->destroy_by_rcu = get_obj("destroy_by_rcu");
 			slab->hwcache_align = get_obj("hwcache_align");
Signature · 1 line
2.20.0.rc2.403.gdbc3b29805-goog
#10 lore raw
In reply to Nicolas Boichat

Hi all,

Quoted 14 lines · Nicolas Boichat

This is a follow-up to the discussion in [1], [2].

IOMMUs using ARMv7 short-descriptor format require page tables (level 1 and 2) to be allocated within the first 4GB of RAM, even on 64-bit systems.

For L1 tables that are bigger than a page, we can just use __get_free_pages with GFP_DMA32 (on arm64 systems only, arm would still use GFP_DMA).

For L2 tables that only take 1KB, it would be a waste to allocate a full page, so we considered 3 approaches:
1. This series, adding support for GFP_DMA32 slab caches.
2. genalloc, which requires pre-allocating the maximum number of L2 page

Quoted 5 lines
    tables (4096, so 4MB of memory).

3. page_frag, which is not very memory-efficient as it is unable to reuse freed fragments until the whole page is freed. [3]

This series is the most memory-efficient approach.

Does anyone have any further comment on this series? If not, which maintainer is going to pick this up? I assume Andrew Morton?

Thanks,

Quoted 56 lines

stable@ note:
We confirmed that this is a regression, and IOMMU errors happen on 4.19 and linux-next/master on MT8173 (elm, Acer Chromebook R13). The issue most likely starts from commit ad67f5a6545f ("arm64: replace ZONE_DMA with ZONE_DMA32"), i.e. 4.15, and presumably breaks a number of Mediatek platforms (and maybe others?).

[1] https://lists.linuxfoundation.org/pipermail/iommu/2018-November/030876.html
[2] https://lists.linuxfoundation.org/pipermail/iommu/2018-December/031696.html
[3] https://patchwork.codeaurora.org/patch/671639/

Changes since v1:
- Add support for SLAB_CACHE_DMA32 in slab and slub (patches 1/2)
- iommu/io-pgtable-arm-v7s (patch 3):
- Changed approach to use SLAB_CACHE_DMA32 added by the previous commit.
- Use DMA or DMA32 depending on the architecture (DMA for arm,

     DMA32 for arm64).

Changes since v2:
- Reworded and expanded commit messages
- Added cache_dma32 documentation in PATCH 2/3.

v3 used the page_frag approach, see [3].

Changes since v4:
- Dropped change that removed GFP_DMA32 from GFP_SLAB_BUG_MASK:
instead we can just call kmem_cache_*alloc without GFP_DMA32 parameter. This also means that we can drop PATCH v4 1/3, as we do not make any changes in GFP flag verification.
- Dropped hunks that added cache_dma32 sysfs file, and moved the hunks to PATCH v5 3/3, so that maintainer can decide whether to pick the change independently.

Changes since v5:
- Rename ARM_V7S_TABLE_SLAB_CACHE to ARM_V7S_TABLE_SLAB_FLAGS.
- Add stable@ to cc.

Nicolas Boichat (3):
mm: Add support for kmem caches in DMA32 zone iommu/io-pgtable-arm-v7s: Request DMA32 memory, and improve debugging
mm: Add /sys/kernel/slab/cache/cache_dma32

 Documentation/ABI/testing/sysfs-kernel-slab |  9 +++++++++
 drivers/iommu/io-pgtable-arm-v7s.c          | 19 +++++++++++++++----
 include/linux/slab.h                        |  2 ++
 mm/slab.c                                   |  2 ++
 mm/slab.h                                   |  3 ++-
 mm/slab_common.c                            |  2 +-
 mm/slub.c                                   | 16 ++++++++++++++++
 tools/vm/slabinfo.c                         |  7 ++++++-
 8 files changed, 53 insertions(+), 7 deletions(-)

--
2.20.0.rc2.403.gdbc3b29805-goog

maintainer
#11 lore raw
Quoted 2 lines · Nicolas Boichat

Does anyone have any further comment on this series? If not, which maintainer is going to pick this up? I assume Andrew Morton?

Probably, yes. I don't like to carry the mm-changes in iommu-tree, so this should go through mm.

Regards,

Joerg

#12 lore raw

Hi Andrew,

Quoted 7 lines · Joerg Roedel
Quoted 2 lines · Nicolas Boichat

Does anyone have any further comment on this series? If not, which maintainer is going to pick this up? I assume Andrew Morton?

Probably, yes. I don't like to carry the mm-changes in iommu-tree, so this should go through mm.

Gentle ping on this series, it seems like it's better if it goes through your tree.

Series still applies cleanly on linux-next, but I'm happy to resend if that helps.

Thanks!

Quoted 3 lines

Regards,

Joerg

#13 lore raw
Quoted 8 lines · Nicolas Boichat

Hi Andrew,

Quoted 5 lines · Joerg Roedel
Quoted 2 lines · Nicolas Boichat

Does anyone have any further comment on this series? If not, which maintainer is going to pick this up? I assume Andrew Morton?

Quoted 8 lines
Quoted 2 lines

Probably, yes. I don't like to carry the mm-changes in iommu-tree, so this should go through mm.

Gentle ping on this series, it seems like it's better if it goes through your tree.

Series still applies cleanly on linux-next, but I'm happy to resend if that helps.

Ping, Andrew?

Quoted 6 lines

Thanks!

Quoted 3 lines

Regards,

Joerg

#14 lore raw
Quoted 13 lines · Vlastimil Babka
Quoted 11 lines · Nicolas Boichat

Hi Andrew,

Quoted 7 lines · Joerg Roedel
Quoted 2 lines · Nicolas Boichat

Does anyone have any further comment on this series? If not, which maintainer is going to pick this up? I assume Andrew Morton?

Probably, yes. I don't like to carry the mm-changes in iommu-tree, so this should go through mm.

Quoted 7 lines
Quoted 5 lines

Gentle ping on this series, it seems like it's better if it goes through your tree.

Series still applies cleanly on linux-next, but I'm happy to resend if that helps.

Ping, Andrew?

Another gentle ping, I still don't see these patches in mmot[ms]. Thanks.

Quoted 7 lines
Quoted 6 lines

Thanks!

Quoted 3 lines

Regards,

Joerg

#15 lore raw
Quoted 18 lines · Nicolas Boichat
Quoted 16 lines · Vlastimil Babka
Quoted 14 lines · Nicolas Boichat

Hi Andrew,

Quoted 7 lines · Joerg Roedel
Quoted 2 lines · Nicolas Boichat

Does anyone have any further comment on this series? If not, which maintainer is going to pick this up? I assume Andrew Morton?

Probably, yes. I don't like to carry the mm-changes in iommu-tree, so this should go through mm.

Gentle ping on this series, it seems like it's better if it goes through your tree.

Quoted 6 lines
Quoted 4 lines
Quoted 2 lines

Series still applies cleanly on linux-next, but I'm happy to resend if that helps.

Ping, Andrew?

Another gentle ping, I still don't see these patches in mmot[ms]. Thanks.

Andrew: AFAICT this still applies cleanly on linux-next/master, so I don't plan to resend... is there any other issues with this series?

This is a regression, so it'd be nice to have it fixed in mainline, eventually.

Thanks,

Quoted 7 lines
Quoted 7 lines
Quoted 6 lines

Thanks!

Quoted 3 lines

Regards,

Joerg

maintainer
#16 lore raw
Quoted 22 lines · On Tue, 19 Mar 2019 15:41:43 +0800 Nicolas Boichat <[email protected]>…
Quoted 21 lines · Nicolas Boichat
Quoted 19 lines · Vlastimil Babka
Quoted 16 lines · Nicolas Boichat

Hi Andrew,

Quoted 7 lines · Joerg Roedel
Quoted 2 lines · Nicolas Boichat

Does anyone have any further comment on this series? If not, which maintainer is going to pick this up? I assume Andrew Morton?

Probably, yes. I don't like to carry the mm-changes in iommu-tree, so this should go through mm.

Gentle ping on this series, it seems like it's better if it goes through your tree.

Series still applies cleanly on linux-next, but I'm happy to resend if that helps.

Quoted 8 lines
Quoted 3 lines
Quoted 1 line

Ping, Andrew?

Another gentle ping, I still don't see these patches in mmot[ms]. Thanks.

Andrew: AFAICT this still applies cleanly on linux-next/master, so I don't plan to resend... is there any other issues with this series?

This is a regression, so it'd be nice to have it fixed in mainline, eventually.

Sorry, seeing "iommu" and "arm" made these escape my gimlet eye.

I'm only seeing acks on [1/3]. What's the review status of [2/3] and [3/3]?

#17 lore raw
Quoted 27 lines · Andrew Morton
Quoted 24 lines · On Tue, 19 Mar 2019 15:41:43 +0800 Nicolas Boichat <[email protected]>…
Quoted 23 lines · Nicolas Boichat
Quoted 20 lines · Vlastimil Babka
Quoted 8 lines · Nicolas Boichat

Hi Andrew,

Quoted 5 lines · Joerg Roedel
Quoted 2 lines · Nicolas Boichat

Does anyone have any further comment on this series? If not, which maintainer is going to pick this up? I assume Andrew Morton?

Quoted 8 lines
Quoted 2 lines

Probably, yes. I don't like to carry the mm-changes in iommu-tree, so this should go through mm.

Gentle ping on this series, it seems like it's better if it goes through your tree.

Series still applies cleanly on linux-next, but I'm happy to resend if that helps.

Ping, Andrew?

Quoted 8 lines
Quoted 6 lines
Quoted 1 line

Another gentle ping, I still don't see these patches in mmot[ms]. Thanks.

Andrew: AFAICT this still applies cleanly on linux-next/master, so I don't plan to resend... is there any other issues with this series?

This is a regression, so it'd be nice to have it fixed in mainline, eventually.

Sorry, seeing "iommu" and "arm" made these escape my gimlet eye.

Thanks for picking them up!

Quoted 1 line

I'm only seeing acks on [1/3]. What's the review status of [2/3] and [3/3]?

Replied on the notification, [2/3] had a Ack, [3/3] is somewhat controversial.