mirror of
https://github.com/fkie-cad/nvd-json-data-feeds.git
synced 2025-06-01 19:21:37 +00:00
25 lines
3.6 KiB
JSON
25 lines
3.6 KiB
JSON
|
{
|
||
|
|
"id": "CVE-2024-50221",
|
||
|
|
"sourceIdentifier": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
|
||
|
|
"published": "2024-11-09T11:15:07.600",
|
||
|
|
"lastModified": "2024-11-09T11:15:07.600",
|
||
|
|
"vulnStatus": "Received",
|
||
|
|
"cveTags": [],
|
||
|
|
"descriptions": [
|
||
|
|
{
|
||
|
|
"lang": "en",
|
||
|
|
"value": "In the Linux kernel, the following vulnerability has been resolved:\n\ndrm/amd/pm: Vangogh: Fix kernel memory out of bounds write\n\nKASAN reports that the GPU metrics table allocated in\nvangogh_tables_init() is not large enough for the memset done in\nsmu_cmn_init_soft_gpu_metrics(). Condensed report follows:\n\n[ 33.861314] BUG: KASAN: slab-out-of-bounds in smu_cmn_init_soft_gpu_metrics+0x73/0x200 [amdgpu]\n[ 33.861799] Write of size 168 at addr ffff888129f59500 by task mangoapp/1067\n...\n[ 33.861808] CPU: 6 UID: 1000 PID: 1067 Comm: mangoapp Tainted: G W 6.12.0-rc4 #356 1a56f59a8b5182eeaf67eb7cb8b13594dd23b544\n[ 33.861816] Tainted: [W]=WARN\n[ 33.861818] Hardware name: Valve Galileo/Galileo, BIOS F7G0107 12/01/2023\n[ 33.861822] Call Trace:\n[ 33.861826] <TASK>\n[ 33.861829] dump_stack_lvl+0x66/0x90\n[ 33.861838] print_report+0xce/0x620\n[ 33.861853] kasan_report+0xda/0x110\n[ 33.862794] kasan_check_range+0xfd/0x1a0\n[ 33.862799] __asan_memset+0x23/0x40\n[ 33.862803] smu_cmn_init_soft_gpu_metrics+0x73/0x200 [amdgpu 13b1bc364ec578808f676eba412c20eaab792779]\n[ 33.863306] vangogh_get_gpu_metrics_v2_4+0x123/0xad0 [amdgpu 13b1bc364ec578808f676eba412c20eaab792779]\n[ 33.864257] vangogh_common_get_gpu_metrics+0xb0c/0xbc0 [amdgpu 13b1bc364ec578808f676eba412c20eaab792779]\n[ 33.865682] amdgpu_dpm_get_gpu_metrics+0xcc/0x110 [amdgpu 13b1bc364ec578808f676eba412c20eaab792779]\n[ 33.866160] amdgpu_get_gpu_metrics+0x154/0x2d0 [amdgpu 13b1bc364ec578808f676eba412c20eaab792779]\n[ 33.867135] dev_attr_show+0x43/0xc0\n[ 33.867147] sysfs_kf_seq_show+0x1f1/0x3b0\n[ 33.867155] seq_read_iter+0x3f8/0x1140\n[ 33.867173] vfs_read+0x76c/0xc50\n[ 33.867198] ksys_read+0xfb/0x1d0\n[ 33.867214] do_syscall_64+0x90/0x160\n...\n[ 33.867353] Allocated by task 378 on cpu 7 at 22.794876s:\n[ 33.867358] kasan_save_stack+0x33/0x50\n[ 33.867364] kasan_save_track+0x17/0x60\n[ 33.867367] __kasan_kmalloc+0x87/0x90\n[ 33.867371] vangogh_init_smc_tables+0x3f9/0x840 [amdgpu]\n[ 33.867835] smu_sw_init+0xa32/0x1850 [amdgpu]\n[ 33.868299] amdgpu_device_init+0x467b/0x8d90 [amdgpu]\n[ 33.868733] amdgpu_driver_load_kms+0x19/0xf0 [amdgpu]\n[ 33.869167] amdgpu_pci_probe+0x2d6/0xcd0 [amdgpu]\n[ 33.869608] local_pci_probe+0xda/0x180\n[ 33.869614] pci_device_probe+0x43f/0x6b0\n\nEmpirically we can confirm that the former allocates 152 bytes for the\ntable, while the latter memsets the 168 large block.\n\nRoot cause appears that when GPU metrics tables for v2_4 parts were added\nit was not considered to enlarge the table to fit.\n\nThe fix in this patch is rather \"brute force\" and perhaps later should be\ndone in a smarter way, by extracting and consolidating the part version to\nsize logic to a common helper, instead of brute forcing the largest\npossible allocation. Nevertheless, for now this works and fixes the out of\nbounds write.\n\nv2:\n * Drop impossible v3_0 case. (Mario)\n\n(cherry picked from commit 0880f58f9609f0200483a49429af0f050d281703)"
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"metrics": {},
|
||
|
|
"references": [
|
||
|
|
{
|
||
|
|
"url": "https://git.kernel.org/stable/c/4aa923a6e6406b43566ef6ac35a3d9a3197fa3e8",
|
||
|
|
"source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"url": "https://git.kernel.org/stable/c/f8fd9f0d57af4f8f48b383ec28287af85b47cb9f",
|
||
|
|
"source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
|
||
|
|
}
|
||
|
|
]
|
||
|
|
}
|