linux-stable.git/kernel/dma, branch linux-6.19.y

dma-debug: Allow multiple invocations of overlapping entries

2026-04-22T11:30:53+00:00

commit eca58535b154e6951327319afda94ac80eae7dc3 upstream.

Repeated DMA mappings with DMA_ATTR_CPU_CACHE_CLEAN trigger the
following splat. This prevents using the attribute in cases where a DMA
region is shared and reused more than seven times.

 ------------[ cut here ]------------
 DMA-API: exceeded 7 overlapping mappings of cacheline 0x000000000438c440
 WARNING: kernel/dma/debug.c:467 at add_dma_entry+0x219/0x280, CPU#4: ibv_rc_pingpong/1644
 Modules linked in: xt_conntrack xt_MASQUERADE nf_conntrack_netlink nfnetlink iptable_nat nf_nat xt_addrtype br_netfilter rpcsec_gss_krb5 auth_rpcgss oid_registry overlay mlx5_fwctl zram zsmalloc mlx5_ib fuse rpcrdma rdma_ucm ib_uverbs ib_iser libiscsi scsi_transport_iscsi ib_umad rdma_cm ib_ipoib iw_cm ib_cm mlx5_core ib_core
 CPU: 4 UID: 2733 PID: 1644 Comm: ibv_rc_pingpong Not tainted 6.19.0+ #129 PREEMPT
 Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014
 RIP: 0010:add_dma_entry+0x221/0x280
 Code: c0 0f 84 f2 fe ff ff 83 e8 01 89 05 6d 99 11 01 e9 e4 fe ff ff 0f 8e 1f ff ff ff 48 8d 3d 07 ef 2d 01 be 07 00 00 00 48 89 e2 <67> 48 0f b9 3a e9 06 ff ff ff 48 c7 c7 98 05 2b 82 c6 05 72 92 28
 RSP: 0018:ff1100010e657970 EFLAGS: 00010002
 RAX: 0000000000000007 RBX: ff1100010234eb00 RCX: 0000000000000000
 RDX: ff1100010e657970 RSI: 0000000000000007 RDI: ffffffff82678660
 RBP: 000000000438c440 R08: 0000000000000228 R09: 0000000000000000
 R10: 00000000000001be R11: 000000000000089d R12: 0000000000000800
 R13: 00000000ffffffef R14: 0000000000000202 R15: ff1100010234eb00
 FS:  00007fb15f3f6740(0000) GS:ff110008dcc19000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 00007fb15f32d3a0 CR3: 0000000116f59001 CR4: 0000000000373eb0
 Call Trace:
  
  debug_dma_map_sg+0x1b4/0x390
  __dma_map_sg_attrs+0x6d/0x1a0
  dma_map_sgtable+0x19/0x30
  ib_umem_get+0x284/0x3b0 [ib_uverbs]
  mlx5_ib_reg_user_mr+0x68/0x2a0 [mlx5_ib]
  ib_uverbs_reg_mr+0x17f/0x2a0 [ib_uverbs]
  ib_uverbs_handler_UVERBS_METHOD_INVOKE_WRITE+0xc2/0x130 [ib_uverbs]
  ib_uverbs_cmd_verbs+0xa0b/0xae0 [ib_uverbs]
  ? ib_uverbs_handler_UVERBS_METHOD_QUERY_PORT_SPEED+0xe0/0xe0 [ib_uverbs]
  ? mmap_region+0x7a/0xb0
  ? do_mmap+0x3b8/0x5c0
  ib_uverbs_ioctl+0xa7/0x110 [ib_uverbs]
  __x64_sys_ioctl+0x14f/0x8b0
  ? ksys_mmap_pgoff+0xc5/0x190
  do_syscall_64+0x8c/0xbf0
  entry_SYSCALL_64_after_hwframe+0x4b/0x53
 RIP: 0033:0x7fb15f5e4eed
 Code: 04 25 28 00 00 00 48 89 45 c8 31 c0 48 8d 45 10 c7 45 b0 10 00 00 00 48 89 45 b8 48 8d 45 d0 48 89 45 c0 b8 10 00 00 00 0f 05 <89> c2 3d 00 f0 ff ff 77 1a 48 8b 45 c8 64 48 2b 04 25 28 00 00 00
 RSP: 002b:00007ffe09a5c540 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
 RAX: ffffffffffffffda RBX: 00007ffe09a5c5d0 RCX: 00007fb15f5e4eed
 RDX: 00007ffe09a5c5f0 RSI: 00000000c0181b01 RDI: 0000000000000003
 RBP: 00007ffe09a5c590 R08: 0000000000000028 R09: 00007ffe09a5c794
 R10: 0000000000000001 R11: 0000000000000246 R12: 00007ffe09a5c794
 R13: 000000000000000c R14: 0000000025a49170 R15: 000000000000000c
  
 ---[ end trace 0000000000000000 ]---

Fixes: 61868dc55a11 ("dma-mapping: add DMA_ATTR_CPU_CACHE_CLEAN")
Signed-off-by: Leon Romanovsky 
Signed-off-by: Marek Szyprowski 
Link: https://lore.kernel.org/r/20260316-dma-debug-overlap-v3-1-1dde90a7f08b@nvidia.com
Signed-off-by: Greg Kroah-Hartman

dma-debug: suppress cacheline overlap warning when arch has no DMA alignment requirement

2026-04-22T11:30:36+00:00

[ Upstream commit 3d48c9fd78dd0b1809669ec49c4d0997b8127512 ]

When CONFIG_DMA_API_DEBUG is enabled, the DMA debug infrastructure
tracks active mappings per cacheline and warns if two different DMA
mappings share the same cacheline ("cacheline tracking EEXIST,
overlapping mappings aren't supported").

On x86_64, ARCH_KMALLOC_MINALIGN defaults to 8, so small kmalloc
allocations (e.g. the 8-byte hub->buffer and hub->status in the USB
hub driver) frequently land in the same 64-byte cacheline.  When both
are DMA-mapped, this triggers a false positive warning.

This has been reported repeatedly since v5.14 (when the EEXIST check
was added) across various USB host controllers and devices including
xhci_hcd with USB hubs, USB audio devices, and USB ethernet adapters.

The cacheline overlap is only a real concern on architectures that
require DMA buffer alignment to cacheline boundaries (i.e. where
ARCH_DMA_MINALIGN >= L1_CACHE_BYTES).  On architectures like x86_64
where dma_get_cache_alignment() returns 1, the hardware is
cache-coherent and overlapping cacheline mappings are harmless.

Suppress the EEXIST warning when dma_get_cache_alignment() is less
than L1_CACHE_BYTES, indicating the architecture does not require
cacheline-aligned DMA buffers.

Verified with a kernel module reproducer that performs two kmalloc(8)
allocations back-to-back and DMA-maps both:

  Before: allocations share a cacheline, EEXIST fires within ~50 pairs
  After:  same cacheline pair found, but no warning emitted

Fixes: 2b4bbc6231d7 ("dma-debug: report -EEXIST errors in add_dma_entry")
Link: https://bugzilla.kernel.org/show_bug.cgi?id=215740
Suggested-by: Harry Yoo 
Tested-by: Mikhail Gavrilov 
Signed-off-by: Mikhail Gavrilov 
Signed-off-by: Marek Szyprowski 
Link: https://lore.kernel.org/r/20260327124156.24820-1-mikhail.v.gavrilov@gmail.com
Signed-off-by: Sasha Levin

dma-debug: track cache clean flag in entries

2026-04-22T11:30:35+00:00

[ Upstream commit d5d846513128c1a3bc2f2d371f6e903177dea443 ]

If a driver is buggy and has 2 overlapping mappings but only
sets cache clean flag on the 1st one of them, we warn.
But if it only does it for the 2nd one, we don't.

Fix by tracking cache clean flag in the entry.

Message-ID: <0ffb3513d18614539c108b4548cdfbc64274a7d1.1767601130.git.mst@redhat.com>
Reviewed-by: Petr Tesarik 
Signed-off-by: Michael S. Tsirkin 
Stable-dep-of: 3d48c9fd78dd ("dma-debug: suppress cacheline overlap warning when arch has no DMA alignment requirement")
Signed-off-by: Sasha Levin

dma-mapping: add DMA_ATTR_CPU_CACHE_CLEAN

2026-04-22T11:30:35+00:00

[ Upstream commit 61868dc55a119a5e4b912d458fc2c48ba80a35fe ]

When multiple small DMA_FROM_DEVICE or DMA_BIDIRECTIONAL buffers share a
cacheline, and DMA_API_DEBUG is enabled, we get this warning:
	cacheline tracking EEXIST, overlapping mappings aren't supported.

This is because when one of the mappings is removed, while another one
is active, CPU might write into the buffer.

Add an attribute for the driver to promise not to do this, making the
overlapping safe, and suppressing the warning.

Message-ID: <2d5d091f9d84b68ea96abd545b365dd1d00bbf48.1767601130.git.mst@redhat.com>
Reviewed-by: Petr Tesarik 
Acked-by: Marek Szyprowski 
Signed-off-by: Michael S. Tsirkin 
Stable-dep-of: 3d48c9fd78dd ("dma-debug: suppress cacheline overlap warning when arch has no DMA alignment requirement")
Signed-off-by: Sasha Levin

dma: swiotlb: add KMSAN annotations to swiotlb_bounce()

2026-04-02T11:25:24+00:00

[ Upstream commit 6f770b73d0311a5b099277653199bb6421c4fed2 ]

When a device performs DMA to a bounce buffer, KMSAN is unaware of
the write and does not mark the data as initialized.  When
swiotlb_bounce() later copies the bounce buffer back to the original
buffer, memcpy propagates the uninitialized shadow to the original
buffer, causing false positive uninit-value reports.

Fix this by calling kmsan_unpoison_memory() on the bounce buffer
before copying it back in the DMA_FROM_DEVICE path, so that memcpy
naturally propagates initialized shadow to the destination.

Suggested-by: Alexander Potapenko 
Link: https://lore.kernel.org/CAG_fn=WUGta-paG1BgsGRoAR+fmuCgh3xo=R3XdzOt_-DqSdHw@mail.gmail.com/
Fixes: 7ade4f10779c ("dma: kmsan: unpoison DMA mappings")
Signed-off-by: Shigeru Yoshida 
Signed-off-by: Marek Szyprowski 
Link: https://lore.kernel.org/r/20260315082750.2375581-1-syoshida@redhat.com
Signed-off-by: Sasha Levin

dma-mapping: avoid random addr value print out on error path

2026-03-04T12:20:53+00:00

[ Upstream commit 47322c469d4a63ac45b705ca83680671ff71c975 ]

dma_addr is unitialized in dma_direct_map_phys() when swiotlb is forced
and DMA_ATTR_MMIO is set which leads to random value print out in
warning. Fix that by just returning DMA_MAPPING_ERROR.

Fixes: e53d29f957b3 ("dma-mapping: convert dma_direct_*map_page to be phys_addr_t based")
Signed-off-by: Jiri Pirko 
Signed-off-by: Marek Szyprowski 
Link: https://lore.kernel.org/r/20260209153809.250835-2-jiri@resnulli.us
Signed-off-by: Sasha Levin

dma: contiguous: Check return value of dma_contiguous_reserve_area()

2026-02-02T08:20:32+00:00

Commit 8f1fc1bf1a3d ("dma: contiguous: Reserve default CMA heap")
introduced a bug where dma_heap_cma_register_heap() is called with
a NULL pointer when dma_contiguous_reserve_area() fails to reserve
the CMA area.

When dma_contiguous_reserve_area() fails, dma_contiguous_default_area
remains NULL (initialized as a global variable), but the code doesn't
check the return value and proceeds to call dma_heap_cma_register_heap()
with this NULL pointer.

Later during boot, add_cma_heaps() iterates through the dma_areas[]
array and attempts to register heaps. When it encounters the NULL
pointer stored by the earlier call, it crashes in __add_cma_heap()
-> dma_heap_add() when trying to dereference the NULL CMA pointer.

The crash manifests as:
  Unable to handle kernel NULL pointer dereference at virtual address
  0000000000000038
  ...
  Call trace:
   dma_heap_add+0x40/0x2b0
   __add_cma_heap+0x80/0xe0
   add_cma_heaps+0x64/0xb0
   do_one_initcall+0x60/0x318
   kernel_init_freeable+0x260/0x2f0
   kernel_init+0x2c/0x168
   ret_from_fork+0x10/0x20

Fix this by checking the return value of dma_contiguous_reserve_area()
and only calling dma_heap_cma_register_heap() when the reservation
succeeds.

Fixes: 8f1fc1bf1a3d ("dma: contiguous: Reserve default CMA heap")
Signed-off-by: Shanker Donthineni 
Reviewed-by: T.J. Mercier 
Signed-off-by: Marek Szyprowski 
Link: https://lore.kernel.org/r/20260129181317.2429196-1-sdonthineni@nvidia.com

dma/pool: distinguish between missing and exhausted atomic pools

2026-01-29T09:23:45+00:00

Currently, dma_alloc_from_pool() unconditionally warns and dumps a stack
trace when an allocation fails, with the message "Failed to get suitable
pool".

This conflates two distinct failure modes:
1. Configuration error: No atomic pool is available for the requested
   DMA mask (a fundamental system setup issue)
2. Resource Exhaustion: A suitable pool exists but is currently full (a
   recoverable runtime state)

This lack of distinction prevents drivers from using __GFP_NOWARN to
suppress error messages during temporary pressure spikes, such as when
awaiting synchronous reclaim of descriptors.

Refactor the error handling to distinguish these cases:
- If no suitable pool is found, keep the unconditional WARN regarding
  the missing pool.
- If a pool was found but is exhausted, respect __GFP_NOWARN and update
  the warning message to explicitly state "DMA pool exhausted".

Fixes: 9420139f516d ("dma-pool: fix coherent pool allocations for IOMMU mappings")
Signed-off-by: Sai Sree Kartheek Adivi 
Reviewed-by: Robin Murphy 
Signed-off-by: Marek Szyprowski 
Link: https://lore.kernel.org/r/20260128133554.3056582-1-s-adivi@ti.com

of: reserved_mem: Allow reserved_mem framework detect "cma=" kernel param

2026-01-28T23:26:36+00:00

When initializing the default cma region, the "cma=" kernel parameter
takes priority over a DT defined linux,cma-default region. Hence, give
the reserved_mem framework the ability to detect this so that the DT
defined cma region can skip initialization accordingly.

Signed-off-by: Oreoluwa Babatunde 
Tested-by: Joy Zou 
Acked-by: Rob Herring (Arm) 
Fixes: 8a6e02d0c00e ("of: reserved_mem: Restructure how the reserved memory regions are processed")
Fixes: 2c223f7239f3 ("of: reserved_mem: Restructure call site for dma_contiguous_early_fixup()")
Link: https://lore.kernel.org/r/20251210002027.1171519-1-oreoluwa.babatunde@oss.qualcomm.com
[mszyprow: rebased onto v6.19-rc1, added fixes tags, added a stub for
 cma_skip_dt_default_reserved_mem() if no CONFIG_DMA_CMA is set]
Signed-off-by: Marek Szyprowski

dma/pool: Avoid allocating redundant pools

2026-01-14T10:00:00+00:00

On smaller systems, e.g. embedded arm64, it is common for all memory
to end up in ZONE_DMA32 or even ZONE_DMA. In such cases it is redundant
to allocate a nominal pool for an empty higher zone that just ends up
coming from a lower zone that should already have its own pool anyway.
We already have logic to skip allocating a ZONE_DMA pool when that is
empty, so generalise that to save memory in the case of other zones too.

Signed-off-by: Robin Murphy 
Tested-by: Vladimir Kondratiev 
Reviewed-by: Baoquan He 
Signed-off-by: Marek Szyprowski 
Link: https://lore.kernel.org/r/8ab8d8a620dee0109f33f5cb63d6bfeed35aac37.1768230104.git.robin.murphy@arm.com