mirror of
https://github.com/fkie-cad/nvd-json-data-feeds.git
synced 2025-07-29 05:56:17 +00:00
24 lines
3.2 KiB
JSON
24 lines
3.2 KiB
JSON
{
|
|
"id": "CVE-2021-47531",
|
|
"sourceIdentifier": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
|
|
"published": "2024-05-24T15:15:16.357",
|
|
"lastModified": "2024-05-24T18:09:20.027",
|
|
"vulnStatus": "Awaiting Analysis",
|
|
"descriptions": [
|
|
{
|
|
"lang": "en",
|
|
"value": "In the Linux kernel, the following vulnerability has been resolved:\n\ndrm/msm: Fix mmap to include VM_IO and VM_DONTDUMP\n\nIn commit 510410bfc034 (\"drm/msm: Implement mmap as GEM object\nfunction\") we switched to a new/cleaner method of doing things. That's\ngood, but we missed a little bit.\n\nBefore that commit, we used to _first_ run through the\ndrm_gem_mmap_obj() case where `obj->funcs->mmap()` was NULL. That meant\nthat we ran:\n\n vma->vm_flags |= VM_IO | VM_PFNMAP | VM_DONTEXPAND | VM_DONTDUMP;\n vma->vm_page_prot = pgprot_writecombine(vm_get_page_prot(vma->vm_flags));\n vma->vm_page_prot = pgprot_decrypted(vma->vm_page_prot);\n\n...and _then_ we modified those mappings with our own. Now that\n`obj->funcs->mmap()` is no longer NULL we don't run the default\ncode. It looks like the fact that the vm_flags got VM_IO / VM_DONTDUMP\nwas important because we're now getting crashes on Chromebooks that\nuse ARC++ while logging out. Specifically a crash that looks like this\n(this is on a 5.10 kernel w/ relevant backports but also seen on a\n5.15 kernel):\n\n Unable to handle kernel paging request at virtual address ffffffc008000000\n Mem abort info:\n ESR = 0x96000006\n EC = 0x25: DABT (current EL), IL = 32 bits\n SET = 0, FnV = 0\n EA = 0, S1PTW = 0\n Data abort info:\n ISV = 0, ISS = 0x00000006\n CM = 0, WnR = 0\n swapper pgtable: 4k pages, 39-bit VAs, pgdp=000000008293d000\n [ffffffc008000000] pgd=00000001002b3003, p4d=00000001002b3003,\n pud=00000001002b3003, pmd=0000000000000000\n Internal error: Oops: 96000006 [#1] PREEMPT SMP\n [...]\n CPU: 7 PID: 15734 Comm: crash_dump64 Tainted: G W 5.10.67 #1 [...]\n Hardware name: Qualcomm Technologies, Inc. sc7280 IDP SKU2 platform (DT)\n pstate: 80400009 (Nzcv daif +PAN -UAO -TCO BTYPE=--)\n pc : __arch_copy_to_user+0xc0/0x30c\n lr : copyout+0xac/0x14c\n [...]\n Call trace:\n __arch_copy_to_user+0xc0/0x30c\n copy_page_to_iter+0x1a0/0x294\n process_vm_rw_core+0x240/0x408\n process_vm_rw+0x110/0x16c\n __arm64_sys_process_vm_readv+0x30/0x3c\n el0_svc_common+0xf8/0x250\n do_el0_svc+0x30/0x80\n el0_svc+0x10/0x1c\n el0_sync_handler+0x78/0x108\n el0_sync+0x184/0x1c0\n Code: f8408423 f80008c3 910020c6 36100082 (b8404423)\n\nLet's add the two flags back in.\n\nWhile we're at it, the fact that we aren't running the default means\nthat we _don't_ need to clear out VM_PFNMAP, so remove that and save\nan instruction.\n\nNOTE: it was confirmed that VM_IO was the important flag to fix the\nproblem I was seeing, but adding back VM_DONTDUMP seems like a sane\nthing to do so I'm doing that too."
|
|
}
|
|
],
|
|
"metrics": {},
|
|
"references": [
|
|
{
|
|
"url": "https://git.kernel.org/stable/c/3466d9e217b337bf473ee629c608e53f9f3ab786",
|
|
"source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
|
|
},
|
|
{
|
|
"url": "https://git.kernel.org/stable/c/8e2b7fe5e8a4be5e571561d9afcfbd92097288ba",
|
|
"source": "416baaa9-dc9f-4396-8d5f-8c081fb06d67"
|
|
}
|
|
]
|
|
} |