mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* KMSAN: uninit-value in eth_type_trans
@ 2025-09-25 22:36 Aleksandr Nogikh
  2025-10-07  6:51 ` Alexander Potapenko
  0 siblings, 1 reply; 6+ messages in thread
From: Aleksandr Nogikh @ 2025-09-25 22:36 UTC (permalink / raw)
  To: anthony.l.nguyen, przemyslaw.kitszel
  Cc: andrew+netdev, davem, edumazet, kuba, pabeni, intel-wired-lan,
	netdev, linux-kernel, syzkaller-bugs, glider

Hello net developers,

I hit the following kernel crash when I try to boot a CONFIG_KMSAN=y kernel on qemu:

KMSAN: uninit-value in eth_type_trans

Could you please have a look?

Kernel: torvalds
Commit: cec1e6e5d1ab33403b809f79cd20d6aff124ccfe
Config: https://raw.githubusercontent.com/google/syzkaller/refs/heads/master/dashboard/config/linux/upstream-kmsan.config

Qemu command to reproduce:

qemu-system-x86_64 -m 8G -smp 2,sockets=2,cores=1 -machine pc-q35-10.0 \
-enable-kvm -display none -serial stdio -snapshot \
-device virtio-blk-pci,drive=myhd -drive file=~/buildroot_amd64_2024.09,format=raw,if=none,id=myhd \
-kernel ~/linux/arch/x86/boot/bzImage -append "root=/dev/vda1" -cpu max \
-net nic,model=e1000 -net user,host=10.0.2.10,hostfwd=tcp:127.0.0.1:10021-:22

The command used the buildroot image below: 
$ wget 'https://storage.googleapis.com/syzkaller/images/buildroot_amd64_2024.09.gz'
$ gunzip buildroot_amd64_2024.09.gz

Full symbolized report:

BUG: KMSAN: uninit-value in eth_skb_pkt_type include/linux/etherdevice.h:627 [inline]
BUG: KMSAN: uninit-value in eth_type_trans+0x4ee/0x980 net/ethernet/eth.c:165
 eth_skb_pkt_type include/linux/etherdevice.h:627 [inline]
 eth_type_trans+0x4ee/0x980 net/ethernet/eth.c:165
 e1000_receive_skb drivers/net/ethernet/intel/e1000/e1000_main.c:4005 [inline]
 e1000_clean_rx_irq+0x1256/0x1cf0 drivers/net/ethernet/intel/e1000/e1000_main.c:4465
 e1000_clean+0x1e4b/0x5f10 drivers/net/ethernet/intel/e1000/e1000_main.c:3807
 __napi_poll+0xda/0x850 net/core/dev.c:7506
 napi_poll net/core/dev.c:7569 [inline]
 net_rx_action+0xa56/0x1b00 net/core/dev.c:7696
 handle_softirqs+0x166/0x6e0 kernel/softirq.c:579
 __do_softirq kernel/softirq.c:613 [inline]
 invoke_softirq kernel/softirq.c:453 [inline]
 __irq_exit_rcu+0x66/0x180 kernel/softirq.c:680
 irq_exit_rcu+0x12/0x20 kernel/softirq.c:696
 common_interrupt+0x99/0xb0 arch/x86/kernel/irq.c:318
 asm_common_interrupt+0x2b/0x40 arch/x86/include/asm/idtentry.h:693
 native_safe_halt arch/x86/include/asm/irqflags.h:48 [inline]
 pv_native_safe_halt+0x17/0x20 arch/x86/kernel/paravirt.c:81
 arch_safe_halt arch/x86/kernel/process.c:756 [inline]
 default_idle+0xd/0x20 arch/x86/kernel/process.c:757
 arch_cpu_idle+0xd/0x20 arch/x86/kernel/process.c:794
 default_idle_call+0x41/0x70 kernel/sched/idle.c:122
 cpuidle_idle_call kernel/sched/idle.c:190 [inline]
 do_idle+0x1dc/0x790 kernel/sched/idle.c:330
 cpu_startup_entry+0x60/0x80 kernel/sched/idle.c:428
 rest_init+0x1df/0x260 init/main.c:744
 start_kernel+0x76e/0x960 init/main.c:1097
 x86_64_start_reservations+0x28/0x30 arch/x86/kernel/head64.c:307
 x86_64_start_kernel+0x139/0x140 arch/x86/kernel/head64.c:288
 common_startup_64+0x13e/0x147

Uninit was stored to memory at:
 skb_put_data include/linux/skbuff.h:2753 [inline]
 e1000_copybreak drivers/net/ethernet/intel/e1000/e1000_main.c:4339 [inline]
 e1000_clean_rx_irq+0x870/0x1cf0 drivers/net/ethernet/intel/e1000/e1000_main.c:4384
 e1000_clean+0x1e4b/0x5f10 drivers/net/ethernet/intel/e1000/e1000_main.c:3807
 __napi_poll+0xda/0x850 net/core/dev.c:7506
 napi_poll net/core/dev.c:7569 [inline]
 net_rx_action+0xa56/0x1b00 net/core/dev.c:7696
 handle_softirqs+0x166/0x6e0 kernel/softirq.c:579
 __do_softirq kernel/softirq.c:613 [inline]
 invoke_softirq kernel/softirq.c:453 [inline]
 __irq_exit_rcu+0x66/0x180 kernel/softirq.c:680
 irq_exit_rcu+0x12/0x20 kernel/softirq.c:696
 common_interrupt+0x99/0xb0 arch/x86/kernel/irq.c:318
 asm_common_interrupt+0x2b/0x40 arch/x86/include/asm/idtentry.h:693

Uninit was stored to memory at:
 swiotlb_bounce+0x470/0x640 kernel/dma/swiotlb.c:-1
 __swiotlb_sync_single_for_cpu+0x9e/0xc0 kernel/dma/swiotlb.c:1567
 swiotlb_sync_single_for_cpu include/linux/swiotlb.h:279 [inline]
 dma_direct_sync_single_for_cpu kernel/dma/direct.h:77 [inline]
 __dma_sync_single_for_cpu+0x50d/0x710 kernel/dma/mapping.c:370
 dma_sync_single_for_cpu include/linux/dma-mapping.h:381 [inline]
 e1000_copybreak drivers/net/ethernet/intel/e1000/e1000_main.c:4336 [inline]
 e1000_clean_rx_irq+0x7dc/0x1cf0 drivers/net/ethernet/intel/e1000/e1000_main.c:4384
 e1000_clean+0x1e4b/0x5f10 drivers/net/ethernet/intel/e1000/e1000_main.c:3807
 __napi_poll+0xda/0x850 net/core/dev.c:7506
 napi_poll net/core/dev.c:7569 [inline]
 net_rx_action+0xa56/0x1b00 net/core/dev.c:7696
 handle_softirqs+0x166/0x6e0 kernel/softirq.c:579
 __do_softirq kernel/softirq.c:613 [inline]
 invoke_softirq kernel/softirq.c:453 [inline]
 __irq_exit_rcu+0x66/0x180 kernel/softirq.c:680
 irq_exit_rcu+0x12/0x20 kernel/softirq.c:696
 common_interrupt+0x99/0xb0 arch/x86/kernel/irq.c:318
 asm_common_interrupt+0x2b/0x40 arch/x86/include/asm/idtentry.h:693

Uninit was stored to memory at:
 swiotlb_bounce+0x470/0x640 kernel/dma/swiotlb.c:-1
 swiotlb_tbl_map_single+0x2956/0x2b20 kernel/dma/swiotlb.c:1439
 swiotlb_map+0x349/0x1050 kernel/dma/swiotlb.c:1584
 dma_direct_map_page kernel/dma/direct.h:-1 [inline]
 dma_map_page_attrs+0x614/0xef0 kernel/dma/mapping.c:169
 dma_map_single_attrs include/linux/dma-mapping.h:469 [inline]
 e1000_alloc_rx_buffers+0x96d/0x1600 drivers/net/ethernet/intel/e1000/e1000_main.c:4616
 e1000_configure+0x16fe/0x1930 drivers/net/ethernet/intel/e1000/e1000_main.c:377
 e1000_open+0x985/0x14d0 drivers/net/ethernet/intel/e1000/e1000_main.c:1388
 __dev_open+0x7c2/0xc40 net/core/dev.c:1682
 __dev_change_flags+0x3ae/0x9b0 net/core/dev.c:9549
 netif_change_flags+0x8d/0x1e0 net/core/dev.c:9612
 dev_change_flags+0x18c/0x320 net/core/dev_api.c:68
 devinet_ioctl+0x162d/0x2570 net/ipv4/devinet.c:1199
 inet_ioctl+0x4c0/0x6f0 net/ipv4/af_inet.c:1001
 sock_do_ioctl+0x9f/0x480 net/socket.c:1238
 sock_ioctl+0x70b/0xd60 net/socket.c:1359
 vfs_ioctl fs/ioctl.c:51 [inline]
 __do_sys_ioctl fs/ioctl.c:598 [inline]
 __se_sys_ioctl+0x23c/0x400 fs/ioctl.c:584
 __x64_sys_ioctl+0x97/0xe0 fs/ioctl.c:584
 x64_sys_call+0x1cbc/0x3e20 arch/x86/include/generated/asm/syscalls_64.h:17
 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
 do_syscall_64+0xd9/0x210 arch/x86/entry/syscall_64.c:94
 entry_SYSCALL_64_after_hwframe+0x77/0x7f

Uninit was created at:
 __alloc_frozen_pages_noprof+0x648/0xe80 mm/page_alloc.c:5171
 __alloc_pages_noprof+0x41/0xd0 mm/page_alloc.c:5182
 __page_frag_cache_refill+0x57/0x2a0 mm/page_frag_cache.c:59
 __page_frag_alloc_align+0xd0/0x690 mm/page_frag_cache.c:103
 __napi_alloc_frag_align net/core/skbuff.c:248 [inline]
 __netdev_alloc_frag_align+0x1b7/0x1f0 net/core/skbuff.c:269
 netdev_alloc_frag include/linux/skbuff.h:3408 [inline]
 e1000_alloc_frag drivers/net/ethernet/intel/e1000/e1000_main.c:2074 [inline]
 e1000_alloc_rx_buffers+0x276/0x1600 drivers/net/ethernet/intel/e1000/e1000_main.c:4584
 e1000_configure+0x16fe/0x1930 drivers/net/ethernet/intel/e1000/e1000_main.c:377
 e1000_open+0x985/0x14d0 drivers/net/ethernet/intel/e1000/e1000_main.c:1388
 __dev_open+0x7c2/0xc40 net/core/dev.c:1682
 __dev_change_flags+0x3ae/0x9b0 net/core/dev.c:9549
 netif_change_flags+0x8d/0x1e0 net/core/dev.c:9612
 dev_change_flags+0x18c/0x320 net/core/dev_api.c:68
 devinet_ioctl+0x162d/0x2570 net/ipv4/devinet.c:1199
 inet_ioctl+0x4c0/0x6f0 net/ipv4/af_inet.c:1001
 sock_do_ioctl+0x9f/0x480 net/socket.c:1238
 sock_ioctl+0x70b/0xd60 net/socket.c:1359
 vfs_ioctl fs/ioctl.c:51 [inline]
 __do_sys_ioctl fs/ioctl.c:598 [inline]
 __se_sys_ioctl+0x23c/0x400 fs/ioctl.c:584
 __x64_sys_ioctl+0x97/0xe0 fs/ioctl.c:584
 x64_sys_call+0x1cbc/0x3e20 arch/x86/include/generated/asm/syscalls_64.h:17
 do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
 do_syscall_64+0xd9/0x210 arch/x86/entry/syscall_64.c:94
 entry_SYSCALL_64_after_hwframe+0x77/0x7f


--
Aleksandr

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: KMSAN: uninit-value in eth_type_trans
  2025-09-25 22:36 KMSAN: uninit-value in eth_type_trans Aleksandr Nogikh
@ 2025-10-07  6:51 ` Alexander Potapenko
  2025-10-08  8:43   ` Alexander Potapenko
  2025-10-08  8:53   ` Marek Szyprowski
  0 siblings, 2 replies; 6+ messages in thread
From: Alexander Potapenko @ 2025-10-07  6:51 UTC (permalink / raw)
  To: Robin Murphy, Christoph Hellwig, Marek Szyprowski,
	Leon Romanovsky, mhklinux
  Cc: anthony.l.nguyen, przemyslaw.kitszel, andrew+netdev, davem,
	edumazet, kuba, pabeni, intel-wired-lan, netdev, linux-kernel,
	syzkaller-bugs, Aleksandr Nogikh

On Fri, Sep 26, 2025 at 12:36 AM Aleksandr Nogikh <nogikh@google.com> wrote:
>
> Hello net developers,

CCing DMA developers, as this seems to be a generic problem.
See the question below, after the KMSAN report.

> I hit the following kernel crash when I try to boot a CONFIG_KMSAN=y kernel on qemu:
>
> KMSAN: uninit-value in eth_type_trans
>
> Could you please have a look?
>
> Kernel: torvalds
> Commit: cec1e6e5d1ab33403b809f79cd20d6aff124ccfe
> Config: https://raw.githubusercontent.com/google/syzkaller/refs/heads/master/dashboard/config/linux/upstream-kmsan.config
>
> Qemu command to reproduce:
>
> qemu-system-x86_64 -m 8G -smp 2,sockets=2,cores=1 -machine pc-q35-10.0 \
> -enable-kvm -display none -serial stdio -snapshot \
> -device virtio-blk-pci,drive=myhd -drive file=~/buildroot_amd64_2024.09,format=raw,if=none,id=myhd \
> -kernel ~/linux/arch/x86/boot/bzImage -append "root=/dev/vda1" -cpu max \
> -net nic,model=e1000 -net user,host=10.0.2.10,hostfwd=tcp:127.0.0.1:10021-:22
>
> The command used the buildroot image below:
> $ wget 'https://storage.googleapis.com/syzkaller/images/buildroot_amd64_2024.09.gz'
> $ gunzip buildroot_amd64_2024.09.gz
>
> Full symbolized report:
>
> BUG: KMSAN: uninit-value in eth_skb_pkt_type include/linux/etherdevice.h:627 [inline]
> BUG: KMSAN: uninit-value in eth_type_trans+0x4ee/0x980 net/ethernet/eth.c:165
>  eth_skb_pkt_type include/linux/etherdevice.h:627 [inline]
>  eth_type_trans+0x4ee/0x980 net/ethernet/eth.c:165
>  e1000_receive_skb drivers/net/ethernet/intel/e1000/e1000_main.c:4005 [inline]
>  e1000_clean_rx_irq+0x1256/0x1cf0 drivers/net/ethernet/intel/e1000/e1000_main.c:4465
>  e1000_clean+0x1e4b/0x5f10 drivers/net/ethernet/intel/e1000/e1000_main.c:3807
>  __napi_poll+0xda/0x850 net/core/dev.c:7506
>  napi_poll net/core/dev.c:7569 [inline]
>  net_rx_action+0xa56/0x1b00 net/core/dev.c:7696
>  handle_softirqs+0x166/0x6e0 kernel/softirq.c:579
>  __do_softirq kernel/softirq.c:613 [inline]
>  invoke_softirq kernel/softirq.c:453 [inline]
>  __irq_exit_rcu+0x66/0x180 kernel/softirq.c:680
>  irq_exit_rcu+0x12/0x20 kernel/softirq.c:696
>  common_interrupt+0x99/0xb0 arch/x86/kernel/irq.c:318
>  asm_common_interrupt+0x2b/0x40 arch/x86/include/asm/idtentry.h:693
>  native_safe_halt arch/x86/include/asm/irqflags.h:48 [inline]
>  pv_native_safe_halt+0x17/0x20 arch/x86/kernel/paravirt.c:81
>  arch_safe_halt arch/x86/kernel/process.c:756 [inline]
>  default_idle+0xd/0x20 arch/x86/kernel/process.c:757
>  arch_cpu_idle+0xd/0x20 arch/x86/kernel/process.c:794
>  default_idle_call+0x41/0x70 kernel/sched/idle.c:122
>  cpuidle_idle_call kernel/sched/idle.c:190 [inline]
>  do_idle+0x1dc/0x790 kernel/sched/idle.c:330
>  cpu_startup_entry+0x60/0x80 kernel/sched/idle.c:428
>  rest_init+0x1df/0x260 init/main.c:744
>  start_kernel+0x76e/0x960 init/main.c:1097
>  x86_64_start_reservations+0x28/0x30 arch/x86/kernel/head64.c:307
>  x86_64_start_kernel+0x139/0x140 arch/x86/kernel/head64.c:288
>  common_startup_64+0x13e/0x147
>
> Uninit was stored to memory at:
>  skb_put_data include/linux/skbuff.h:2753 [inline]
>  e1000_copybreak drivers/net/ethernet/intel/e1000/e1000_main.c:4339 [inline]
>  e1000_clean_rx_irq+0x870/0x1cf0 drivers/net/ethernet/intel/e1000/e1000_main.c:4384
>  e1000_clean+0x1e4b/0x5f10 drivers/net/ethernet/intel/e1000/e1000_main.c:3807
>  __napi_poll+0xda/0x850 net/core/dev.c:7506
>  napi_poll net/core/dev.c:7569 [inline]
>  net_rx_action+0xa56/0x1b00 net/core/dev.c:7696
>  handle_softirqs+0x166/0x6e0 kernel/softirq.c:579
>  __do_softirq kernel/softirq.c:613 [inline]
>  invoke_softirq kernel/softirq.c:453 [inline]
>  __irq_exit_rcu+0x66/0x180 kernel/softirq.c:680
>  irq_exit_rcu+0x12/0x20 kernel/softirq.c:696
>  common_interrupt+0x99/0xb0 arch/x86/kernel/irq.c:318
>  asm_common_interrupt+0x2b/0x40 arch/x86/include/asm/idtentry.h:693
>
> Uninit was stored to memory at:
>  swiotlb_bounce+0x470/0x640 kernel/dma/swiotlb.c:-1
>  __swiotlb_sync_single_for_cpu+0x9e/0xc0 kernel/dma/swiotlb.c:1567
>  swiotlb_sync_single_for_cpu include/linux/swiotlb.h:279 [inline]
>  dma_direct_sync_single_for_cpu kernel/dma/direct.h:77 [inline]
>  __dma_sync_single_for_cpu+0x50d/0x710 kernel/dma/mapping.c:370
>  dma_sync_single_for_cpu include/linux/dma-mapping.h:381 [inline]
>  e1000_copybreak drivers/net/ethernet/intel/e1000/e1000_main.c:4336 [inline]
>  e1000_clean_rx_irq+0x7dc/0x1cf0 drivers/net/ethernet/intel/e1000/e1000_main.c:4384
>  e1000_clean+0x1e4b/0x5f10 drivers/net/ethernet/intel/e1000/e1000_main.c:3807
>  __napi_poll+0xda/0x850 net/core/dev.c:7506
>  napi_poll net/core/dev.c:7569 [inline]
>  net_rx_action+0xa56/0x1b00 net/core/dev.c:7696
>  handle_softirqs+0x166/0x6e0 kernel/softirq.c:579
>  __do_softirq kernel/softirq.c:613 [inline]
>  invoke_softirq kernel/softirq.c:453 [inline]
>  __irq_exit_rcu+0x66/0x180 kernel/softirq.c:680
>  irq_exit_rcu+0x12/0x20 kernel/softirq.c:696
>  common_interrupt+0x99/0xb0 arch/x86/kernel/irq.c:318
>  asm_common_interrupt+0x2b/0x40 arch/x86/include/asm/idtentry.h:693
>
> Uninit was stored to memory at:
>  swiotlb_bounce+0x470/0x640 kernel/dma/swiotlb.c:-1
>  swiotlb_tbl_map_single+0x2956/0x2b20 kernel/dma/swiotlb.c:1439
>  swiotlb_map+0x349/0x1050 kernel/dma/swiotlb.c:1584
>  dma_direct_map_page kernel/dma/direct.h:-1 [inline]
>  dma_map_page_attrs+0x614/0xef0 kernel/dma/mapping.c:169
>  dma_map_single_attrs include/linux/dma-mapping.h:469 [inline]
>  e1000_alloc_rx_buffers+0x96d/0x1600 drivers/net/ethernet/intel/e1000/e1000_main.c:4616
>  e1000_configure+0x16fe/0x1930 drivers/net/ethernet/intel/e1000/e1000_main.c:377
>  e1000_open+0x985/0x14d0 drivers/net/ethernet/intel/e1000/e1000_main.c:1388
>  __dev_open+0x7c2/0xc40 net/core/dev.c:1682
>  __dev_change_flags+0x3ae/0x9b0 net/core/dev.c:9549
>  netif_change_flags+0x8d/0x1e0 net/core/dev.c:9612
>  dev_change_flags+0x18c/0x320 net/core/dev_api.c:68
>  devinet_ioctl+0x162d/0x2570 net/ipv4/devinet.c:1199
>  inet_ioctl+0x4c0/0x6f0 net/ipv4/af_inet.c:1001
>  sock_do_ioctl+0x9f/0x480 net/socket.c:1238
>  sock_ioctl+0x70b/0xd60 net/socket.c:1359
>  vfs_ioctl fs/ioctl.c:51 [inline]
>  __do_sys_ioctl fs/ioctl.c:598 [inline]
>  __se_sys_ioctl+0x23c/0x400 fs/ioctl.c:584
>  __x64_sys_ioctl+0x97/0xe0 fs/ioctl.c:584
>  x64_sys_call+0x1cbc/0x3e20 arch/x86/include/generated/asm/syscalls_64.h:17
>  do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
>  do_syscall_64+0xd9/0x210 arch/x86/entry/syscall_64.c:94
>  entry_SYSCALL_64_after_hwframe+0x77/0x7f
>
> Uninit was created at:
>  __alloc_frozen_pages_noprof+0x648/0xe80 mm/page_alloc.c:5171
>  __alloc_pages_noprof+0x41/0xd0 mm/page_alloc.c:5182
>  __page_frag_cache_refill+0x57/0x2a0 mm/page_frag_cache.c:59
>  __page_frag_alloc_align+0xd0/0x690 mm/page_frag_cache.c:103
>  __napi_alloc_frag_align net/core/skbuff.c:248 [inline]
>  __netdev_alloc_frag_align+0x1b7/0x1f0 net/core/skbuff.c:269
>  netdev_alloc_frag include/linux/skbuff.h:3408 [inline]
>  e1000_alloc_frag drivers/net/ethernet/intel/e1000/e1000_main.c:2074 [inline]
>  e1000_alloc_rx_buffers+0x276/0x1600 drivers/net/ethernet/intel/e1000/e1000_main.c:4584
>  e1000_configure+0x16fe/0x1930 drivers/net/ethernet/intel/e1000/e1000_main.c:377
>  e1000_open+0x985/0x14d0 drivers/net/ethernet/intel/e1000/e1000_main.c:1388
>  __dev_open+0x7c2/0xc40 net/core/dev.c:1682
>  __dev_change_flags+0x3ae/0x9b0 net/core/dev.c:9549
>  netif_change_flags+0x8d/0x1e0 net/core/dev.c:9612
>  dev_change_flags+0x18c/0x320 net/core/dev_api.c:68
>  devinet_ioctl+0x162d/0x2570 net/ipv4/devinet.c:1199
>  inet_ioctl+0x4c0/0x6f0 net/ipv4/af_inet.c:1001
>  sock_do_ioctl+0x9f/0x480 net/socket.c:1238
>  sock_ioctl+0x70b/0xd60 net/socket.c:1359
>  vfs_ioctl fs/ioctl.c:51 [inline]
>  __do_sys_ioctl fs/ioctl.c:598 [inline]
>  __se_sys_ioctl+0x23c/0x400 fs/ioctl.c:584
>  __x64_sys_ioctl+0x97/0xe0 fs/ioctl.c:584
>  x64_sys_call+0x1cbc/0x3e20 arch/x86/include/generated/asm/syscalls_64.h:17
>  do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
>  do_syscall_64+0xd9/0x210 arch/x86/entry/syscall_64.c:94
>  entry_SYSCALL_64_after_hwframe+0x77/0x7f

Folks, as far as I understand, dma_direct_sync_single_for_cpu() and
dma_direct_sync_single_for_device() are the places where we send data
to or from the device.
Should we add KMSAN annotations to those functions to catch infoleaks
and mark data from devices as initialized?

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: KMSAN: uninit-value in eth_type_trans
  2025-10-07  6:51 ` Alexander Potapenko
@ 2025-10-08  8:43   ` Alexander Potapenko
  2025-10-13  7:31     ` Marek Szyprowski
  2025-10-08  8:53   ` Marek Szyprowski
  1 sibling, 1 reply; 6+ messages in thread
From: Alexander Potapenko @ 2025-10-08  8:43 UTC (permalink / raw)
  To: Robin Murphy, Christoph Hellwig, Marek Szyprowski,
	Leon Romanovsky, mhklinux
  Cc: anthony.l.nguyen, przemyslaw.kitszel, andrew+netdev, davem,
	edumazet, kuba, pabeni, netdev, linux-kernel, syzkaller-bugs,
	Aleksandr Nogikh

On Tue, Oct 7, 2025 at 8:51 AM Alexander Potapenko <glider@google.com> wrote:
> Folks, as far as I understand, dma_direct_sync_single_for_cpu() and
> dma_direct_sync_single_for_device() are the places where we send data
> to or from the device.
> Should we add KMSAN annotations to those functions to catch infoleaks
> and mark data from devices as initialized?

Something along the lines of:

======================================
diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c
index 0d37da3d95b65..7f59de19c1c87 100644
--- a/kernel/dma/swiotlb.c
+++ b/kernel/dma/swiotlb.c
@@ -42,6 +42,7 @@
 #include <linux/string.h>
 #include <linux/swiotlb.h>
 #include <linux/types.h>
+#include <linux/kmsan-checks.h>
 #ifdef CONFIG_DMA_RESTRICTED_POOL
 #include <linux/of.h>
 #include <linux/of_fdt.h>
@@ -903,10 +904,13 @@ static void swiotlb_bounce(struct device *dev,
phys_addr_t tlb_addr, size_t size

                        local_irq_save(flags);
                        page = pfn_to_page(pfn);
-                       if (dir == DMA_TO_DEVICE)
+                       if (dir == DMA_TO_DEVICE) {
+                               kmsan_check_highmem_page(page, offset, sz);
                                memcpy_from_page(vaddr, page, offset, sz);
-                       else
+                       } else {
+                               kmsan_unpoison_memory(vaddr, sz);
                                memcpy_to_page(page, offset, vaddr, sz);
+                       }
                        local_irq_restore(flags);

                        size -= sz;
@@ -915,8 +919,10 @@ static void swiotlb_bounce(struct device *dev,
phys_addr_t tlb_addr, size_t size
                        offset = 0;
                }
        } else if (dir == DMA_TO_DEVICE) {
+               kmsan_check_memory(phys_to_virt(orig_addr), size);
                memcpy(vaddr, phys_to_virt(orig_addr), size);
        } else {
+               kmsan_unpoison_memory(vaddr, size);
                memcpy(phys_to_virt(orig_addr), vaddr, size);
        }
 }
======================================

should be conceptually right, but according to the comment in
swiotlb_tbl_map_single()
(https://elixir.bootlin.com/linux/v6.17.1/source/kernel/dma/swiotlb.c#L1431),
that function is deliberately copying the buffer to the device, even
when it is uninitialized - and KMSAN actually started reporting that
when I applied the above patch.

How should we handle this case?
Not adding the kmsan_check_memory() calls will solve the problem, but
there might be real infoleaks that we won't detect.
We could unpoison the buffer before passing it to
swiotlb_tbl_map_single() to ignore just the first infoleak on the
buffer.
Alternatively, we could require callers to always initialize the
buffer passed to swiotlb_tbl_map_single().

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: KMSAN: uninit-value in eth_type_trans
  2025-10-07  6:51 ` Alexander Potapenko
  2025-10-08  8:43   ` Alexander Potapenko
@ 2025-10-08  8:53   ` Marek Szyprowski
  1 sibling, 0 replies; 6+ messages in thread
From: Marek Szyprowski @ 2025-10-08  8:53 UTC (permalink / raw)
  To: Alexander Potapenko, Robin Murphy, Christoph Hellwig,
	Leon Romanovsky, mhklinux
  Cc: anthony.l.nguyen, przemyslaw.kitszel, andrew+netdev, davem,
	edumazet, kuba, pabeni, intel-wired-lan, netdev, linux-kernel,
	syzkaller-bugs, Aleksandr Nogikh, iommu

Hi

On 07.10.2025 08:51, Alexander Potapenko wrote:
> On Fri, Sep 26, 2025 at 12:36 AM Aleksandr Nogikh <nogikh@google.com> wrote:
>> Hello net developers,
> CCing DMA developers, as this seems to be a generic problem.
> See the question below, after the KMSAN report.

Thanks for this report!

>> I hit the following kernel crash when I try to boot a CONFIG_KMSAN=y kernel on qemu:
>>
>> KMSAN: uninit-value in eth_type_trans
>>
>> Could you please have a look?
>>
>> Kernel: torvalds
>> Commit: cec1e6e5d1ab33403b809f79cd20d6aff124ccfe
>> Config: https://protect2.fireeye.com/v1/url?k=53229ed1-0cb9a7ce-5323159e-000babdfecba-dfefb775995a4321&q=1&e=a1f2777b-91b5-43b3-8580-011d41f4d75a&u=https%3A%2F%2Fraw.githubusercontent.com%2Fgoogle%2Fsyzkaller%2Frefs%2Fheads%2Fmaster%2Fdashboard%2Fconfig%2Flinux%2Fupstream-kmsan.config
>>
>> Qemu command to reproduce:
>>
>> qemu-system-x86_64 -m 8G -smp 2,sockets=2,cores=1 -machine pc-q35-10.0 \
>> -enable-kvm -display none -serial stdio -snapshot \
>> -device virtio-blk-pci,drive=myhd -drive file=~/buildroot_amd64_2024.09,format=raw,if=none,id=myhd \
>> -kernel ~/linux/arch/x86/boot/bzImage -append "root=/dev/vda1" -cpu max \
>> -net nic,model=e1000 -net user,host=10.0.2.10,hostfwd=tcp:127.0.0.1:10021-:22
>>
>> The command used the buildroot image below:
>> $ wget 'https://protect2.fireeye.com/v1/url?k=b0fed9f8-ef65e0e7-b0ff52b7-000babdfecba-ea62e0c7e52f3737&q=1&e=a1f2777b-91b5-43b3-8580-011d41f4d75a&u=https%3A%2F%2Fstorage.googleapis.com%2Fsyzkaller%2Fimages%2Fbuildroot_amd64_2024.09.gz'
>> $ gunzip buildroot_amd64_2024.09.gz
>>
>> Full symbolized report:
>>
>> BUG: KMSAN: uninit-value in eth_skb_pkt_type include/linux/etherdevice.h:627 [inline]
>> BUG: KMSAN: uninit-value in eth_type_trans+0x4ee/0x980 net/ethernet/eth.c:165
>>   eth_skb_pkt_type include/linux/etherdevice.h:627 [inline]
>>   eth_type_trans+0x4ee/0x980 net/ethernet/eth.c:165
>>   e1000_receive_skb drivers/net/ethernet/intel/e1000/e1000_main.c:4005 [inline]
>>   e1000_clean_rx_irq+0x1256/0x1cf0 drivers/net/ethernet/intel/e1000/e1000_main.c:4465
>>   e1000_clean+0x1e4b/0x5f10 drivers/net/ethernet/intel/e1000/e1000_main.c:3807
>>   __napi_poll+0xda/0x850 net/core/dev.c:7506
>>   napi_poll net/core/dev.c:7569 [inline]
>>   net_rx_action+0xa56/0x1b00 net/core/dev.c:7696
>>   handle_softirqs+0x166/0x6e0 kernel/softirq.c:579
>>   __do_softirq kernel/softirq.c:613 [inline]
>>   invoke_softirq kernel/softirq.c:453 [inline]
>>   __irq_exit_rcu+0x66/0x180 kernel/softirq.c:680
>>   irq_exit_rcu+0x12/0x20 kernel/softirq.c:696
>>   common_interrupt+0x99/0xb0 arch/x86/kernel/irq.c:318
>>   asm_common_interrupt+0x2b/0x40 arch/x86/include/asm/idtentry.h:693
>>   native_safe_halt arch/x86/include/asm/irqflags.h:48 [inline]
>>   pv_native_safe_halt+0x17/0x20 arch/x86/kernel/paravirt.c:81
>>   arch_safe_halt arch/x86/kernel/process.c:756 [inline]
>>   default_idle+0xd/0x20 arch/x86/kernel/process.c:757
>>   arch_cpu_idle+0xd/0x20 arch/x86/kernel/process.c:794
>>   default_idle_call+0x41/0x70 kernel/sched/idle.c:122
>>   cpuidle_idle_call kernel/sched/idle.c:190 [inline]
>>   do_idle+0x1dc/0x790 kernel/sched/idle.c:330
>>   cpu_startup_entry+0x60/0x80 kernel/sched/idle.c:428
>>   rest_init+0x1df/0x260 init/main.c:744
>>   start_kernel+0x76e/0x960 init/main.c:1097
>>   x86_64_start_reservations+0x28/0x30 arch/x86/kernel/head64.c:307
>>   x86_64_start_kernel+0x139/0x140 arch/x86/kernel/head64.c:288
>>   common_startup_64+0x13e/0x147
>>
>> Uninit was stored to memory at:
>>   skb_put_data include/linux/skbuff.h:2753 [inline]
>>   e1000_copybreak drivers/net/ethernet/intel/e1000/e1000_main.c:4339 [inline]
>>   e1000_clean_rx_irq+0x870/0x1cf0 drivers/net/ethernet/intel/e1000/e1000_main.c:4384
>>   e1000_clean+0x1e4b/0x5f10 drivers/net/ethernet/intel/e1000/e1000_main.c:3807
>>   __napi_poll+0xda/0x850 net/core/dev.c:7506
>>   napi_poll net/core/dev.c:7569 [inline]
>>   net_rx_action+0xa56/0x1b00 net/core/dev.c:7696
>>   handle_softirqs+0x166/0x6e0 kernel/softirq.c:579
>>   __do_softirq kernel/softirq.c:613 [inline]
>>   invoke_softirq kernel/softirq.c:453 [inline]
>>   __irq_exit_rcu+0x66/0x180 kernel/softirq.c:680
>>   irq_exit_rcu+0x12/0x20 kernel/softirq.c:696
>>   common_interrupt+0x99/0xb0 arch/x86/kernel/irq.c:318
>>   asm_common_interrupt+0x2b/0x40 arch/x86/include/asm/idtentry.h:693
>>
>> Uninit was stored to memory at:
>>   swiotlb_bounce+0x470/0x640 kernel/dma/swiotlb.c:-1
>>   __swiotlb_sync_single_for_cpu+0x9e/0xc0 kernel/dma/swiotlb.c:1567
>>   swiotlb_sync_single_for_cpu include/linux/swiotlb.h:279 [inline]
>>   dma_direct_sync_single_for_cpu kernel/dma/direct.h:77 [inline]
>>   __dma_sync_single_for_cpu+0x50d/0x710 kernel/dma/mapping.c:370
>>   dma_sync_single_for_cpu include/linux/dma-mapping.h:381 [inline]
>>   e1000_copybreak drivers/net/ethernet/intel/e1000/e1000_main.c:4336 [inline]
>>   e1000_clean_rx_irq+0x7dc/0x1cf0 drivers/net/ethernet/intel/e1000/e1000_main.c:4384
>>   e1000_clean+0x1e4b/0x5f10 drivers/net/ethernet/intel/e1000/e1000_main.c:3807
>>   __napi_poll+0xda/0x850 net/core/dev.c:7506
>>   napi_poll net/core/dev.c:7569 [inline]
>>   net_rx_action+0xa56/0x1b00 net/core/dev.c:7696
>>   handle_softirqs+0x166/0x6e0 kernel/softirq.c:579
>>   __do_softirq kernel/softirq.c:613 [inline]
>>   invoke_softirq kernel/softirq.c:453 [inline]
>>   __irq_exit_rcu+0x66/0x180 kernel/softirq.c:680
>>   irq_exit_rcu+0x12/0x20 kernel/softirq.c:696
>>   common_interrupt+0x99/0xb0 arch/x86/kernel/irq.c:318
>>   asm_common_interrupt+0x2b/0x40 arch/x86/include/asm/idtentry.h:693
>>
>> Uninit was stored to memory at:
>>   swiotlb_bounce+0x470/0x640 kernel/dma/swiotlb.c:-1
>>   swiotlb_tbl_map_single+0x2956/0x2b20 kernel/dma/swiotlb.c:1439
>>   swiotlb_map+0x349/0x1050 kernel/dma/swiotlb.c:1584
>>   dma_direct_map_page kernel/dma/direct.h:-1 [inline]
>>   dma_map_page_attrs+0x614/0xef0 kernel/dma/mapping.c:169
>>   dma_map_single_attrs include/linux/dma-mapping.h:469 [inline]
>>   e1000_alloc_rx_buffers+0x96d/0x1600 drivers/net/ethernet/intel/e1000/e1000_main.c:4616
>>   e1000_configure+0x16fe/0x1930 drivers/net/ethernet/intel/e1000/e1000_main.c:377
>>   e1000_open+0x985/0x14d0 drivers/net/ethernet/intel/e1000/e1000_main.c:1388
>>   __dev_open+0x7c2/0xc40 net/core/dev.c:1682
>>   __dev_change_flags+0x3ae/0x9b0 net/core/dev.c:9549
>>   netif_change_flags+0x8d/0x1e0 net/core/dev.c:9612
>>   dev_change_flags+0x18c/0x320 net/core/dev_api.c:68
>>   devinet_ioctl+0x162d/0x2570 net/ipv4/devinet.c:1199
>>   inet_ioctl+0x4c0/0x6f0 net/ipv4/af_inet.c:1001
>>   sock_do_ioctl+0x9f/0x480 net/socket.c:1238
>>   sock_ioctl+0x70b/0xd60 net/socket.c:1359
>>   vfs_ioctl fs/ioctl.c:51 [inline]
>>   __do_sys_ioctl fs/ioctl.c:598 [inline]
>>   __se_sys_ioctl+0x23c/0x400 fs/ioctl.c:584
>>   __x64_sys_ioctl+0x97/0xe0 fs/ioctl.c:584
>>   x64_sys_call+0x1cbc/0x3e20 arch/x86/include/generated/asm/syscalls_64.h:17
>>   do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
>>   do_syscall_64+0xd9/0x210 arch/x86/entry/syscall_64.c:94
>>   entry_SYSCALL_64_after_hwframe+0x77/0x7f
>>
>> Uninit was created at:
>>   __alloc_frozen_pages_noprof+0x648/0xe80 mm/page_alloc.c:5171
>>   __alloc_pages_noprof+0x41/0xd0 mm/page_alloc.c:5182
>>   __page_frag_cache_refill+0x57/0x2a0 mm/page_frag_cache.c:59
>>   __page_frag_alloc_align+0xd0/0x690 mm/page_frag_cache.c:103
>>   __napi_alloc_frag_align net/core/skbuff.c:248 [inline]
>>   __netdev_alloc_frag_align+0x1b7/0x1f0 net/core/skbuff.c:269
>>   netdev_alloc_frag include/linux/skbuff.h:3408 [inline]
>>   e1000_alloc_frag drivers/net/ethernet/intel/e1000/e1000_main.c:2074 [inline]
>>   e1000_alloc_rx_buffers+0x276/0x1600 drivers/net/ethernet/intel/e1000/e1000_main.c:4584
>>   e1000_configure+0x16fe/0x1930 drivers/net/ethernet/intel/e1000/e1000_main.c:377
>>   e1000_open+0x985/0x14d0 drivers/net/ethernet/intel/e1000/e1000_main.c:1388
>>   __dev_open+0x7c2/0xc40 net/core/dev.c:1682
>>   __dev_change_flags+0x3ae/0x9b0 net/core/dev.c:9549
>>   netif_change_flags+0x8d/0x1e0 net/core/dev.c:9612
>>   dev_change_flags+0x18c/0x320 net/core/dev_api.c:68
>>   devinet_ioctl+0x162d/0x2570 net/ipv4/devinet.c:1199
>>   inet_ioctl+0x4c0/0x6f0 net/ipv4/af_inet.c:1001
>>   sock_do_ioctl+0x9f/0x480 net/socket.c:1238
>>   sock_ioctl+0x70b/0xd60 net/socket.c:1359
>>   vfs_ioctl fs/ioctl.c:51 [inline]
>>   __do_sys_ioctl fs/ioctl.c:598 [inline]
>>   __se_sys_ioctl+0x23c/0x400 fs/ioctl.c:584
>>   __x64_sys_ioctl+0x97/0xe0 fs/ioctl.c:584
>>   x64_sys_call+0x1cbc/0x3e20 arch/x86/include/generated/asm/syscalls_64.h:17
>>   do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
>>   do_syscall_64+0xd9/0x210 arch/x86/entry/syscall_64.c:94
>>   entry_SYSCALL_64_after_hwframe+0x77/0x7f
> Folks, as far as I understand, dma_direct_sync_single_for_cpu() and
> dma_direct_sync_single_for_device() are the places where we send data
> to or from the device.
> Should we add KMSAN annotations to those functions to catch infoleaks
> and mark data from devices as initialized?

I confirm the issue and indeed dma_sync* function family requires kmsan 
annotations. Those should be added in the same place as trace_dma_* and 
debug_dma_* calls in kernel/dma/mapping.c. I briefly looked at the 
existing annotations there and found that the existing 
kmsan_handle_dma() calls also should be moved from dma_map* to 
dma_unmap* set of functions, because only after them it is safe to 
access the DMA transferred data by the CPU.

The major problem however is that in dma_unmap_page() (or 
dma_unmap_phys() in linus/master) and __dma_sync_single*() there is no 
access to original page pointer needed by kmsan hook. The only way to 
fix this is probably to add .dma_to_phys() method to struct dma_map_ops 
and all its providers.


I made a quick PoC based on dma-direct and it resolved the issue 
reported in this thread on QEMU system:

diff --git a/kernel/dma/mapping.c b/kernel/dma/mapping.c
index 56de28a3b179..98ba1a6b5c84 100644
--- a/kernel/dma/mapping.c
+++ b/kernel/dma/mapping.c
@@ -171,7 +171,6 @@ dma_addr_t dma_map_page_attrs(struct device *dev, 
struct page *page,
                 addr = iommu_dma_map_page(dev, page, offset, size, dir, 
attrs);
         else
                 addr = ops->map_page(dev, page, offset, size, dir, attrs);
-       kmsan_handle_dma(page, offset, size, dir);
         trace_dma_map_page(dev, page_to_phys(page) + offset, addr, 
size, dir,
                            attrs);
         debug_dma_map_page(dev, page, offset, size, dir, addr, attrs);
@@ -180,12 +179,30 @@ dma_addr_t dma_map_page_attrs(struct device *dev, 
struct page *page,
  }
  EXPORT_SYMBOL(dma_map_page_attrs);

+#include <linux/dma-direct.h>
+
+static void kmsan_handle_direct_dma(struct device *dev, dma_addr_t addr,
+               size_t size, enum dma_data_direction dir)
+{
+       phys_addr_t phys = dma_to_phys(dev, addr);
+       struct page *page = pfn_to_page(PHYS_PFN(phys));
+       size_t offset = offset_in_page(phys);
+
+       kmsan_handle_dma(page, offset, size, dir);
+}
+
  void dma_unmap_page_attrs(struct device *dev, dma_addr_t addr, size_t 
size,
                 enum dma_data_direction dir, unsigned long attrs)
  {
         const struct dma_map_ops *ops = get_dma_ops(dev);

         BUG_ON(!valid_dma_direction(dir));
+
+       if (dma_map_direct(dev, ops))
+               kmsan_handle_direct_dma(dev, addr, size, dir);
+       else
+               BUG();
+
         if (dma_map_direct(dev, ops) ||
             arch_dma_unmap_page_direct(dev, addr + size))
                 dma_direct_unmap_page(dev, addr, size, dir, attrs);
@@ -218,7 +235,6 @@ static int __dma_map_sg_attrs(struct device *dev, 
struct scatterlist *sg,
                 ents = ops->map_sg(dev, sg, nents, dir, attrs);

         if (ents > 0) {
-               kmsan_handle_dma_sg(sg, nents, dir);
                 trace_dma_map_sg(dev, sg, nents, ents, dir, attrs);
                 debug_dma_map_sg(dev, sg, nents, ents, dir, attrs);
         } else if (WARN_ON_ONCE(ents != -EINVAL && ents != -ENOMEM &&
@@ -306,6 +322,7 @@ void dma_unmap_sg_attrs(struct device *dev, struct 
scatterlist *sg,
         const struct dma_map_ops *ops = get_dma_ops(dev);

         BUG_ON(!valid_dma_direction(dir));
+       kmsan_handle_dma_sg(sg, nents, dir);
         trace_dma_unmap_sg(dev, sg, nents, dir, attrs);
         debug_dma_unmap_sg(dev, sg, nents, dir);
         if (dma_map_direct(dev, ops) ||
@@ -366,6 +383,12 @@ void __dma_sync_single_for_cpu(struct device *dev, 
dma_addr_t addr, size_t size,
         const struct dma_map_ops *ops = get_dma_ops(dev);

         BUG_ON(!valid_dma_direction(dir));
+
+       if (dma_map_direct(dev, ops))
+               kmsan_handle_direct_dma(dev, addr, size, dir);
+       else
+               BUG();
+
         if (dma_map_direct(dev, ops))
                 dma_direct_sync_single_for_cpu(dev, addr, size, dir);
         else if (use_dma_iommu(dev))
@@ -406,6 +429,7 @@ void __dma_sync_sg_for_cpu(struct device *dev, 
struct scatterlist *sg,
                 iommu_dma_sync_sg_for_cpu(dev, sg, nelems, dir);
         else if (ops->sync_sg_for_cpu)
                 ops->sync_sg_for_cpu(dev, sg, nelems, dir);
+       kmsan_handle_dma_sg(sg, nelems, dir);
         trace_dma_sync_sg_for_cpu(dev, sg, nelems, dir);
         debug_dma_sync_sg_for_cpu(dev, sg, nelems, dir);
  }

Best regards
-- 
Marek Szyprowski, PhD
Samsung R&D Institute Poland


^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: KMSAN: uninit-value in eth_type_trans
  2025-10-08  8:43   ` Alexander Potapenko
@ 2025-10-13  7:31     ` Marek Szyprowski
  2026-03-20 17:30       ` Helen Koike
  0 siblings, 1 reply; 6+ messages in thread
From: Marek Szyprowski @ 2025-10-13  7:31 UTC (permalink / raw)
  To: Alexander Potapenko, Robin Murphy, Christoph Hellwig,
	Leon Romanovsky, mhklinux
  Cc: anthony.l.nguyen, przemyslaw.kitszel, andrew+netdev, davem,
	edumazet, kuba, pabeni, netdev, linux-kernel, syzkaller-bugs,
	Aleksandr Nogikh

On 08.10.2025 10:43, Alexander Potapenko wrote:
> On Tue, Oct 7, 2025 at 8:51 AM Alexander Potapenko <glider@google.com> wrote:
>> Folks, as far as I understand, dma_direct_sync_single_for_cpu() and
>> dma_direct_sync_single_for_device() are the places where we send data
>> to or from the device.
>> Should we add KMSAN annotations to those functions to catch infoleaks
>> and mark data from devices as initialized?
> Something along the lines of:
>
> ======================================
> diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c
> index 0d37da3d95b65..7f59de19c1c87 100644
> --- a/kernel/dma/swiotlb.c
> +++ b/kernel/dma/swiotlb.c
> @@ -42,6 +42,7 @@
>   #include <linux/string.h>
>   #include <linux/swiotlb.h>
>   #include <linux/types.h>
> +#include <linux/kmsan-checks.h>
>   #ifdef CONFIG_DMA_RESTRICTED_POOL
>   #include <linux/of.h>
>   #include <linux/of_fdt.h>
> @@ -903,10 +904,13 @@ static void swiotlb_bounce(struct device *dev,
> phys_addr_t tlb_addr, size_t size
>
>                          local_irq_save(flags);
>                          page = pfn_to_page(pfn);
> -                       if (dir == DMA_TO_DEVICE)
> +                       if (dir == DMA_TO_DEVICE) {
> +                               kmsan_check_highmem_page(page, offset, sz);
>                                  memcpy_from_page(vaddr, page, offset, sz);
> -                       else
> +                       } else {
> +                               kmsan_unpoison_memory(vaddr, sz);
>                                  memcpy_to_page(page, offset, vaddr, sz);
> +                       }
>                          local_irq_restore(flags);
>
>                          size -= sz;
> @@ -915,8 +919,10 @@ static void swiotlb_bounce(struct device *dev,
> phys_addr_t tlb_addr, size_t size
>                          offset = 0;
>                  }
>          } else if (dir == DMA_TO_DEVICE) {
> +               kmsan_check_memory(phys_to_virt(orig_addr), size);
>                  memcpy(vaddr, phys_to_virt(orig_addr), size);
>          } else {
> +               kmsan_unpoison_memory(vaddr, size);
>                  memcpy(phys_to_virt(orig_addr), vaddr, size);
>          }
>   }
> ======================================
>
> should be conceptually right, but according to the comment in
> swiotlb_tbl_map_single()
> (https://protect2.fireeye.com/v1/url?k=837a6d67-dce15478-837be628-000babdfecba-aa25926458f9fd30&q=1&e=a3963b1b-328b-4f69-8ca5-ffd6fc777dd7&u=https%3A%2F%2Felixir.bootlin.com%2Flinux%2Fv6.17.1%2Fsource%2Fkernel%2Fdma%2Fswiotlb.c%23L1431),
> that function is deliberately copying the buffer to the device, even
> when it is uninitialized - and KMSAN actually started reporting that
> when I applied the above patch.
>
> How should we handle this case?
> Not adding the kmsan_check_memory() calls will solve the problem, but
> there might be real infoleaks that we won't detect.
> We could unpoison the buffer before passing it to
> swiotlb_tbl_map_single() to ignore just the first infoleak on the
> buffer.
> Alternatively, we could require callers to always initialize the
> buffer passed to swiotlb_tbl_map_single().

Well, I didn't consider swiotlb a special case so far. I did a simple 
test with my PoC patch mentioned earlier in this thread with 
'swiotlb=force' kernel parameter and I didn't observe any kmsan issues, 
but I admin that this wasn't exhaustive test.

Best regards
-- 
Marek Szyprowski, PhD
Samsung R&D Institute Poland


^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: KMSAN: uninit-value in eth_type_trans
  2025-10-13  7:31     ` Marek Szyprowski
@ 2026-03-20 17:30       ` Helen Koike
  0 siblings, 0 replies; 6+ messages in thread
From: Helen Koike @ 2026-03-20 17:30 UTC (permalink / raw)
  To: Marek Szyprowski, Alexander Potapenko, Robin Murphy,
	Christoph Hellwig, Leon Romanovsky, mhklinux
  Cc: anthony.l.nguyen, przemyslaw.kitszel, andrew+netdev, davem,
	edumazet, kuba, pabeni, netdev, linux-kernel, syzkaller-bugs,
	Aleksandr Nogikh, kernel-dev

Hi all,

On 10/13/25 4:31 AM, Marek Szyprowski wrote:
> On 08.10.2025 10:43, Alexander Potapenko wrote:
>> On Tue, Oct 7, 2025 at 8:51 AM Alexander Potapenko <glider@google.com> wrote:
>>> Folks, as far as I understand, dma_direct_sync_single_for_cpu() and
>>> dma_direct_sync_single_for_device() are the places where we send data
>>> to or from the device.
>>> Should we add KMSAN annotations to those functions to catch infoleaks
>>> and mark data from devices as initialized?
>> Something along the lines of:
>>
>> ======================================
>> diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c
>> index 0d37da3d95b65..7f59de19c1c87 100644
>> --- a/kernel/dma/swiotlb.c
>> +++ b/kernel/dma/swiotlb.c
>> @@ -42,6 +42,7 @@
>>    #include <linux/string.h>
>>    #include <linux/swiotlb.h>
>>    #include <linux/types.h>
>> +#include <linux/kmsan-checks.h>
>>    #ifdef CONFIG_DMA_RESTRICTED_POOL
>>    #include <linux/of.h>
>>    #include <linux/of_fdt.h>
>> @@ -903,10 +904,13 @@ static void swiotlb_bounce(struct device *dev,
>> phys_addr_t tlb_addr, size_t size
>>
>>                           local_irq_save(flags);
>>                           page = pfn_to_page(pfn);
>> -                       if (dir == DMA_TO_DEVICE)
>> +                       if (dir == DMA_TO_DEVICE) {
>> +                               kmsan_check_highmem_page(page, offset, sz);
>>                                   memcpy_from_page(vaddr, page, offset, sz);
>> -                       else
>> +                       } else {
>> +                               kmsan_unpoison_memory(vaddr, sz);
>>                                   memcpy_to_page(page, offset, vaddr, sz);
>> +                       }
>>                           local_irq_restore(flags);
>>
>>                           size -= sz;
>> @@ -915,8 +919,10 @@ static void swiotlb_bounce(struct device *dev,
>> phys_addr_t tlb_addr, size_t size
>>                           offset = 0;
>>                   }
>>           } else if (dir == DMA_TO_DEVICE) {
>> +               kmsan_check_memory(phys_to_virt(orig_addr), size);
>>                   memcpy(vaddr, phys_to_virt(orig_addr), size);
>>           } else {
>> +               kmsan_unpoison_memory(vaddr, size);
>>                   memcpy(phys_to_virt(orig_addr), vaddr, size);
>>           }
>>    }
>> ======================================
>>
>> should be conceptually right, but according to the comment in
>> swiotlb_tbl_map_single()
>> (https://protect2.fireeye.com/v1/url?k=837a6d67-dce15478-837be628-000babdfecba-aa25926458f9fd30&q=1&e=a3963b1b-328b-4f69-8ca5-ffd6fc777dd7&u=https%3A%2F%2Felixir.bootlin.com%2Flinux%2Fv6.17.1%2Fsource%2Fkernel%2Fdma%2Fswiotlb.c%23L1431),
>> that function is deliberately copying the buffer to the device, even
>> when it is uninitialized - and KMSAN actually started reporting that
>> when I applied the above patch.
>>
>> How should we handle this case?
>> Not adding the kmsan_check_memory() calls will solve the problem, but
>> there might be real infoleaks that we won't detect.
>> We could unpoison the buffer before passing it to
>> swiotlb_tbl_map_single() to ignore just the first infoleak on the
>> buffer.
>> Alternatively, we could require callers to always initialize the
>> buffer passed to swiotlb_tbl_map_single().
> 
> Well, I didn't consider swiotlb a special case so far. I did a simple
> test with my PoC patch mentioned earlier in this thread with
> 'swiotlb=force' kernel parameter and I didn't observe any kmsan issues,
> but I admin that this wasn't exhaustive test.
> 
> Best regards


I just hit this issue and found this thread. So I'm sharing below my 
report in case it is useful (if not, please just ignore):

I'm easily reproducing it when trying to reproduce another unrelated 
issue[1] by syzbot on qemu[2] with torvalds/master branch, commit 
0e4f8f1a3d08 .

[1] https://syzkaller.appspot.com/bug?extid=ba80855313e6fa65717a
[2] 
https://github.com/google/syzkaller/blob/master/docs/reproducing_crashes.md#using-a-c-reproducer


To momentarily fix it (and allow me o reach my syzbot issue), I just did 
the following:

@@ -916,6 +917,12 @@ static void swiotlb_bounce(struct device *dev, 
phys_addr_t tlb_addr, size_t size
                 memcpy(vaddr, phys_to_virt(orig_addr), size);
         } else {
                 memcpy(phys_to_virt(orig_addr), vaddr, size);
+               kmsan_unpoison_memory(phys_to_virt(orig_addr), size);
         }
  }

Which, from the discussions here is obviously not the right solution.


In case it helps, see below what I'm executing:
(copy and pasted from my scripts)

* Config: 
https://syzkaller.appspot.com/text?tag=KernelConfig&x=963de479f54c6dbb
* Disk image: 
https://storage.googleapis.com/syzbot-assets/bef78207d3f7/disk-2d1373e4.raw.xz


$ gcc \
     repro.c \
     -lpthread \
     -static \
     -o repro

Running make olddefconfig …

$ make O=/$BUILD_DIR CC=clang LD=ld.lld olddefconfig

Building kernel …

$ make O=$BUILD_DIR CC=clang LD=ld.lld -j$NPROC

Starting QEMU with disk image …
SSH port: 35053

$ qemu-system-x86_64 \
     -m 8G \
     -smp 2,sockets=2,cores=1 \
     -drive file=disk.raw,format=raw \
     -net nic,model=e1000 \
     -net user,host=10.0.2.10,hostfwd=tcp::10022-:22 \
     -kernel $BUILD_DIR/arch/x86/boot/bzImage \
     -append 'root=/dev/sda1 console=ttyS0 panic_on_warn=1 
earlyprintk=serial net.ifnames=0 vsyscall=native numa=fake=2' \
     -enable-kvm \
     -nographic \
     -snapshot \
     -machine pc-q35-7.1 \
     -cpu host

$ scp \
     -P 10022 \
     repro \
     root@127.0.0.1:/root/

$ ssh -p 10022 root@127.0.0.1 'chmod +x ./repro && ./repro'


Regards,
Helen


^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2026-03-20 17:31 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2025-09-25 22:36 KMSAN: uninit-value in eth_type_trans Aleksandr Nogikh
2025-10-07  6:51 ` Alexander Potapenko
2025-10-08  8:43   ` Alexander Potapenko
2025-10-13  7:31     ` Marek Szyprowski
2026-03-20 17:30       ` Helen Koike
2025-10-08  8:53   ` Marek Szyprowski

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®