CVE-2021-46984
Vulnerability from cvelistv5
Published
2024-02-28 08:13
Modified
2024-11-04 11:57
Severity ?
EPSS score ?
Summary
kyber: fix out of bounds access when preempted
References
{
"containers": {
"adp": [
{
"providerMetadata": {
"dateUpdated": "2024-08-04T05:24:37.898Z",
"orgId": "af854a3a-2127-422b-91ae-364da2661108",
"shortName": "CVE"
},
"references": [
{
"tags": [
"x_transferred"
],
"url": "https://git.kernel.org/stable/c/0b6b4b90b74c27bea968c214d820ba4254b903a5"
},
{
"tags": [
"x_transferred"
],
"url": "https://git.kernel.org/stable/c/54dbe2d2c1fcabf650c7a8b747601da355cd7f9f"
},
{
"tags": [
"x_transferred"
],
"url": "https://git.kernel.org/stable/c/a287cd84e047045f5a4d4da793414e848de627c6"
},
{
"tags": [
"x_transferred"
],
"url": "https://git.kernel.org/stable/c/2ef3c76540c49167a0bc3d5f80d00fd1fc4586df"
},
{
"tags": [
"x_transferred"
],
"url": "https://git.kernel.org/stable/c/efed9a3337e341bd0989161b97453b52567bc59d"
}
],
"title": "CVE Program Container"
},
{
"metrics": [
{
"other": {
"content": {
"id": "CVE-2021-46984",
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "no"
},
{
"Technical Impact": "partial"
}
],
"role": "CISA Coordinator",
"timestamp": "2024-09-10T16:01:11.596982Z",
"version": "2.0.3"
},
"type": "ssvc"
}
}
],
"providerMetadata": {
"dateUpdated": "2024-09-11T17:33:40.792Z",
"orgId": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
"shortName": "CISA-ADP"
},
"title": "CISA ADP Vulnrichment"
}
],
"cna": {
"affected": [
{
"defaultStatus": "unaffected",
"product": "Linux",
"programFiles": [
"block/bfq-iosched.c",
"block/blk-mq-sched.c",
"block/kyber-iosched.c",
"block/mq-deadline.c",
"include/linux/elevator.h"
],
"repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
"vendor": "Linux",
"versions": [
{
"lessThan": "0b6b4b90b74c",
"status": "affected",
"version": "a6088845c2bf",
"versionType": "git"
},
{
"lessThan": "54dbe2d2c1fc",
"status": "affected",
"version": "a6088845c2bf",
"versionType": "git"
},
{
"lessThan": "a287cd84e047",
"status": "affected",
"version": "a6088845c2bf",
"versionType": "git"
},
{
"lessThan": "2ef3c76540c4",
"status": "affected",
"version": "a6088845c2bf",
"versionType": "git"
},
{
"lessThan": "efed9a3337e3",
"status": "affected",
"version": "a6088845c2bf",
"versionType": "git"
}
]
},
{
"defaultStatus": "affected",
"product": "Linux",
"programFiles": [
"block/bfq-iosched.c",
"block/blk-mq-sched.c",
"block/kyber-iosched.c",
"block/mq-deadline.c",
"include/linux/elevator.h"
],
"repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
"vendor": "Linux",
"versions": [
{
"status": "affected",
"version": "4.18"
},
{
"lessThan": "4.18",
"status": "unaffected",
"version": "0",
"versionType": "semver"
},
{
"lessThanOrEqual": "5.4.*",
"status": "unaffected",
"version": "5.4.120",
"versionType": "semver"
},
{
"lessThanOrEqual": "5.10.*",
"status": "unaffected",
"version": "5.10.38",
"versionType": "semver"
},
{
"lessThanOrEqual": "5.11.*",
"status": "unaffected",
"version": "5.11.22",
"versionType": "semver"
},
{
"lessThanOrEqual": "5.12.*",
"status": "unaffected",
"version": "5.12.5",
"versionType": "semver"
},
{
"lessThanOrEqual": "*",
"status": "unaffected",
"version": "5.13",
"versionType": "original_commit_for_fix"
}
]
}
],
"descriptions": [
{
"lang": "en",
"value": "In the Linux kernel, the following vulnerability has been resolved:\n\nkyber: fix out of bounds access when preempted\n\n__blk_mq_sched_bio_merge() gets the ctx and hctx for the current CPU and\npasses the hctx to -\u003ebio_merge(). kyber_bio_merge() then gets the ctx\nfor the current CPU again and uses that to get the corresponding Kyber\ncontext in the passed hctx. However, the thread may be preempted between\nthe two calls to blk_mq_get_ctx(), and the ctx returned the second time\nmay no longer correspond to the passed hctx. This \"works\" accidentally\nmost of the time, but it can cause us to read garbage if the second ctx\ncame from an hctx with more ctx\u0027s than the first one (i.e., if\nctx-\u003eindex_hw[hctx-\u003etype] \u003e hctx-\u003enr_ctx).\n\nThis manifested as this UBSAN array index out of bounds error reported\nby Jakub:\n\nUBSAN: array-index-out-of-bounds in ../kernel/locking/qspinlock.c:130:9\nindex 13106 is out of range for type \u0027long unsigned int [128]\u0027\nCall Trace:\n dump_stack+0xa4/0xe5\n ubsan_epilogue+0x5/0x40\n __ubsan_handle_out_of_bounds.cold.13+0x2a/0x34\n queued_spin_lock_slowpath+0x476/0x480\n do_raw_spin_lock+0x1c2/0x1d0\n kyber_bio_merge+0x112/0x180\n blk_mq_submit_bio+0x1f5/0x1100\n submit_bio_noacct+0x7b0/0x870\n submit_bio+0xc2/0x3a0\n btrfs_map_bio+0x4f0/0x9d0\n btrfs_submit_data_bio+0x24e/0x310\n submit_one_bio+0x7f/0xb0\n submit_extent_page+0xc4/0x440\n __extent_writepage_io+0x2b8/0x5e0\n __extent_writepage+0x28d/0x6e0\n extent_write_cache_pages+0x4d7/0x7a0\n extent_writepages+0xa2/0x110\n do_writepages+0x8f/0x180\n __writeback_single_inode+0x99/0x7f0\n writeback_sb_inodes+0x34e/0x790\n __writeback_inodes_wb+0x9e/0x120\n wb_writeback+0x4d2/0x660\n wb_workfn+0x64d/0xa10\n process_one_work+0x53a/0xa80\n worker_thread+0x69/0x5b0\n kthread+0x20b/0x240\n ret_from_fork+0x1f/0x30\n\nOnly Kyber uses the hctx, so fix it by passing the request_queue to\n-\u003ebio_merge() instead. BFQ and mq-deadline just use that, and Kyber can\nmap the queues itself to avoid the mismatch."
}
],
"providerMetadata": {
"dateUpdated": "2024-11-04T11:57:08.351Z",
"orgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
"shortName": "Linux"
},
"references": [
{
"url": "https://git.kernel.org/stable/c/0b6b4b90b74c27bea968c214d820ba4254b903a5"
},
{
"url": "https://git.kernel.org/stable/c/54dbe2d2c1fcabf650c7a8b747601da355cd7f9f"
},
{
"url": "https://git.kernel.org/stable/c/a287cd84e047045f5a4d4da793414e848de627c6"
},
{
"url": "https://git.kernel.org/stable/c/2ef3c76540c49167a0bc3d5f80d00fd1fc4586df"
},
{
"url": "https://git.kernel.org/stable/c/efed9a3337e341bd0989161b97453b52567bc59d"
}
],
"title": "kyber: fix out of bounds access when preempted",
"x_generator": {
"engine": "bippy-9e1c9544281a"
}
}
},
"cveMetadata": {
"assignerOrgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
"assignerShortName": "Linux",
"cveId": "CVE-2021-46984",
"datePublished": "2024-02-28T08:13:12.835Z",
"dateReserved": "2024-02-27T18:42:55.946Z",
"dateUpdated": "2024-11-04T11:57:08.351Z",
"state": "PUBLISHED"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.1",
"meta": {
"nvd": "{\"cve\":{\"id\":\"CVE-2021-46984\",\"sourceIdentifier\":\"416baaa9-dc9f-4396-8d5f-8c081fb06d67\",\"published\":\"2024-02-28T09:15:37.450\",\"lastModified\":\"2024-02-28T14:06:45.783\",\"vulnStatus\":\"Awaiting Analysis\",\"descriptions\":[{\"lang\":\"en\",\"value\":\"In the Linux kernel, the following vulnerability has been resolved:\\n\\nkyber: fix out of bounds access when preempted\\n\\n__blk_mq_sched_bio_merge() gets the ctx and hctx for the current CPU and\\npasses the hctx to -\u003ebio_merge(). kyber_bio_merge() then gets the ctx\\nfor the current CPU again and uses that to get the corresponding Kyber\\ncontext in the passed hctx. However, the thread may be preempted between\\nthe two calls to blk_mq_get_ctx(), and the ctx returned the second time\\nmay no longer correspond to the passed hctx. This \\\"works\\\" accidentally\\nmost of the time, but it can cause us to read garbage if the second ctx\\ncame from an hctx with more ctx\u0027s than the first one (i.e., if\\nctx-\u003eindex_hw[hctx-\u003etype] \u003e hctx-\u003enr_ctx).\\n\\nThis manifested as this UBSAN array index out of bounds error reported\\nby Jakub:\\n\\nUBSAN: array-index-out-of-bounds in ../kernel/locking/qspinlock.c:130:9\\nindex 13106 is out of range for type \u0027long unsigned int [128]\u0027\\nCall Trace:\\n dump_stack+0xa4/0xe5\\n ubsan_epilogue+0x5/0x40\\n __ubsan_handle_out_of_bounds.cold.13+0x2a/0x34\\n queued_spin_lock_slowpath+0x476/0x480\\n do_raw_spin_lock+0x1c2/0x1d0\\n kyber_bio_merge+0x112/0x180\\n blk_mq_submit_bio+0x1f5/0x1100\\n submit_bio_noacct+0x7b0/0x870\\n submit_bio+0xc2/0x3a0\\n btrfs_map_bio+0x4f0/0x9d0\\n btrfs_submit_data_bio+0x24e/0x310\\n submit_one_bio+0x7f/0xb0\\n submit_extent_page+0xc4/0x440\\n __extent_writepage_io+0x2b8/0x5e0\\n __extent_writepage+0x28d/0x6e0\\n extent_write_cache_pages+0x4d7/0x7a0\\n extent_writepages+0xa2/0x110\\n do_writepages+0x8f/0x180\\n __writeback_single_inode+0x99/0x7f0\\n writeback_sb_inodes+0x34e/0x790\\n __writeback_inodes_wb+0x9e/0x120\\n wb_writeback+0x4d2/0x660\\n wb_workfn+0x64d/0xa10\\n process_one_work+0x53a/0xa80\\n worker_thread+0x69/0x5b0\\n kthread+0x20b/0x240\\n ret_from_fork+0x1f/0x30\\n\\nOnly Kyber uses the hctx, so fix it by passing the request_queue to\\n-\u003ebio_merge() instead. BFQ and mq-deadline just use that, and Kyber can\\nmap the queues itself to avoid the mismatch.\"},{\"lang\":\"es\",\"value\":\"En el kernel de Linux, se resolvi\u00f3 la siguiente vulnerabilidad: kyber: corrige el acceso fuera de los l\u00edmites cuando se adelanta __blk_mq_sched_bio_merge() obtiene el ctx y el hctx para la CPU actual y pasa el hctx a -\u0026gt;bio_merge(). kyber_bio_merge() luego obtiene nuevamente el ctx para la CPU actual y lo usa para obtener el contexto Kyber correspondiente en el hctx pasado. Sin embargo, el hilo puede ser reemplazado entre las dos llamadas a blk_mq_get_ctx(), y es posible que el ctx devuelto la segunda vez ya no corresponda al hctx pasado. Esto \\\"funciona\\\" accidentalmente la mayor parte del tiempo, pero puede hacer que leamos basura si el segundo ctx proviene de un hctx con m\u00e1s ctx que el primero (es decir, si ctx-\u0026gt;index_hw[hctx-\u0026gt;type] \u0026gt; hctx- \u0026gt;nr_ctx). Esto se manifest\u00f3 como este error de \u00edndice de matriz UBSAN fuera de los l\u00edmites informado por Jakub: UBSAN: array-index-out-of-bounds in ../kernel/locking/qspinlock.c:130:9 index 13106 is out of range for type \u0027 long unsigned int [128]\u0027 Seguimiento de llamadas: dump_stack+0xa4/0xe5 ubsan_epilogue+0x5/0x40 __ubsan_handle_out_of_bounds.cold.13+0x2a/0x34 queued_spin_lock_slowpath+0x476/0x480 do_raw_spin_lock+0x1c2/0x1d0 kyber_bio_ fusionar+0x112/0x180 blk_mq_submit_bio+0x1f5/0x1100 submit_bio_noacct +0x7b0/0x870 submit_bio+0xc2/0x3a0 btrfs_map_bio+0x4f0/0x9d0 btrfs_submit_data_bio+0x24e/0x310 submit_one_bio+0x7f/0xb0 submit_extent_page+0xc4/0x440 __extent_writepage_io+0x2b8/0x5e0 __ extensi\u00f3n_writepage+0x28d/0x6e0 extensi\u00f3n_write_cache_pages+0x4d7/0x7a0 extensi\u00f3n_writepages+0xa2/0x110 do_writepages +0x8f/0x180 __writeback_single_inode+0x99/0x7f0 writeback_sb_inodes+0x34e/0x790 __writeback_inodes_wb+0x9e/0x120 wb_writeback+0x4d2/0x660 wb_workfn+0x64d/0xa10 Process_one_work+0x53a/0xa80 work_thread+0x69/0x5b0 kthread+0x20b/0x240 ret_from_fork+0x1f/0x30 solamente Kyber usa hctx, as\u00ed que solucionelo pasando request_queue a -\u0026gt;bio_merge() en su lugar. BFQ y mq-deadline simplemente usan eso, y Kyber puede mapear las colas \u00e9l mismo para evitar discrepancias.\"}],\"metrics\":{},\"references\":[{\"url\":\"https://git.kernel.org/stable/c/0b6b4b90b74c27bea968c214d820ba4254b903a5\",\"source\":\"416baaa9-dc9f-4396-8d5f-8c081fb06d67\"},{\"url\":\"https://git.kernel.org/stable/c/2ef3c76540c49167a0bc3d5f80d00fd1fc4586df\",\"source\":\"416baaa9-dc9f-4396-8d5f-8c081fb06d67\"},{\"url\":\"https://git.kernel.org/stable/c/54dbe2d2c1fcabf650c7a8b747601da355cd7f9f\",\"source\":\"416baaa9-dc9f-4396-8d5f-8c081fb06d67\"},{\"url\":\"https://git.kernel.org/stable/c/a287cd84e047045f5a4d4da793414e848de627c6\",\"source\":\"416baaa9-dc9f-4396-8d5f-8c081fb06d67\"},{\"url\":\"https://git.kernel.org/stable/c/efed9a3337e341bd0989161b97453b52567bc59d\",\"source\":\"416baaa9-dc9f-4396-8d5f-8c081fb06d67\"}]}}"
}
}