Bug 2240561 (CVE-2023-5158)
| Summary: | CVE-2023-5158 kernel: Possible DOS from guest to host invringh_kiov_advance in vhost driver at drivers/vhost/vringh.c | ||
|---|---|---|---|
| Product: | [Other] Security Response | Reporter: | Rohit Keshri <rkeshri> |
| Component: | vulnerability | Assignee: | Product Security <prodsec-ir-bot> |
| Status: | NEW --- | QA Contact: | |
| Severity: | medium | Docs Contact: | |
| Priority: | medium | ||
| Version: | unspecified | CC: | acaringi, allarkin, bhu, carnil, chwhite, dbohanno, debarbos, dfreiber, dvlasenk, ezulian, hkrzesin, jarod, jburrell, jdenham, jfaracco, jforbes, jlelli, joe.lawrence, jshortt, jstancek, jwyatt, kcarcia, ldoskova, lgoncalv, lzampier, nmurray, ptalbert, rogbas, rrobaina, rvrbovsk, rysulliv, scweaver, security-response-team, sgarzare, tglozar, tyberry, vgoyal, vkumar, walters, wcosta, williams, wmealing, ycote |
| Target Milestone: | --- | Keywords: | Security |
| Target Release: | --- | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | kernel 5.13-rc1 | Doc Type: | If docs needed, set a value |
| Doc Text: |
A flaw was found in vringh_kiov_advance in drivers/vhost/vringh.c in the host side of a virtio ring in the Linux Kernel. This issue may result in a denial of service from guest to host via zero length descriptor.
|
Story Points: | --- |
| Clone Of: | Environment: | ||
| Last Closed: | Type: | --- | |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
| Bug Depends On: | |||
| Bug Blocks: | 2240260 | ||
(In reply to Rohit Keshri from comment #0) > > The fix is to revert b8c06ad4d67db56ed6bdfb685c134da74e92a2c7 ("vringh: > implement vringh_kiov_advance()"). We use vringh_kiov_advance() also in vdpa-sim-blk, so we can't just revert that patch. So I just posted a patch reverting only the changes in vringh_iov_xfer() that should fix the issue reported in this BZ: https://lore.kernel.org/virtualization/20230925103057.104541-1-sgarzare@redhat.com/T/#u The Bugzilla entry mentions "Fixed In Version: kernel 5.13-rc1" but that should rather be that this version is introducing the issue. In reply to comment #7: > The Bugzilla entry mentions "Fixed In Version: kernel 5.13-rc1" but that > should rather be that this version is introducing the issue. Thanks Carnil for correction. $ git describe --contains b8c06ad4d67db56ed6bdfb685c134da74e92a2c7 v5.13-rc1~42^2~14 This is the patch that possibly introduces the DOS. Patch mentioned in comment 6 now merged upstream and in stable branches: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=7aed44babc7f97e82b38e9a68515e699692cc100 |
A possible DOS from guest to host via zero length descriptor. This is because currently the vringh_kiov_advance() did something like this: while (len && iov->i < iov->used) { ... if (!iov->iov[iov->i].iov_len) { ... iov->i++; } } When the length of the descriptor is zero, we miss the chance to advance iov->i. The fix is to revert b8c06ad4d67db56ed6bdfb685c134da74e92a2c7 ("vringh: implement vringh_kiov_advance()").