Bug 2265652 (CVE-2023-52447)

Summary:	CVE-2023-52447 kernel: bpf: Defer the free of inner map when necessary
Product:	[Other] Security Response	Reporter:	Patrick Del Bello <pdelbell>
Component:	vulnerability	Assignee:	Product Security <prodsec-ir-bot>
Status:	NEW ---	QA Contact:
Severity:	low	Docs Contact:
Priority:	low
Version:	unspecified	CC:	acaringi, allarkin, aquini, bhu, chwhite, cye, cyin, dbohanno, debarbos, dfreiber, drow, dvlasenk, esandeen, ezulian, hkrzesin, jarod, jburrell, jdenham, jfaracco, jforbes, jlelli, joe.lawrence, jshortt, jstancek, jwyatt, kcarcia, ldoskova, lgoncalv, lzampier, mleitner, mmilgram, mstowell, nmurray, ptalbert, rparrazo, rrobaina, rvrbovsk, rysulliv, scweaver, sukulkar, tglozar, tyberry, vinair, vkumar, wcosta, williams, wmealing, ycote, ykopkova, zhijwang
Target Milestone:	---	Keywords:	Security
Target Release:	---	Flags:	allarkin: needinfo? (vinair)
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		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:	2265670
Bug Blocks:	2265643

Description Patrick Del Bello 2024-02-23 13:47:34 UTC

bpf: Defer the free of inner map when necessary

When updating or deleting an inner map in map array or map htab, the map
may still be accessed by non-sleepable program or sleepable program.
However bpf_map_fd_put_ptr() decreases the ref-counter of the inner map
directly through bpf_map_put(), if the ref-counter is the last one
(which is true for most cases), the inner map will be freed by
ops->map_free() in a kworker. But for now, most .map_free() callbacks
don't use synchronize_rcu() or its variants to wait for the elapse of a
RCU grace period, so after the invocation of ops->map_free completes,
the bpf program which is accessing the inner map may incur
use-after-free problem.

Fix the free of inner map by invoking bpf_map_free_deferred() after both
one RCU grace period and one tasks trace RCU grace period if the inner
map has been removed from the outer map before. The deferment is
accomplished by using call_rcu() or call_rcu_tasks_trace() when
releasing the last ref-counter of bpf map. The newly-added rcu_head
field in bpf_map shares the same storage space with work field to
reduce the size of bpf_map.

Comment 1 Patrick Del Bello 2024-02-23 14:56:00 UTC

Created kernel tracking bugs for this issue:

Affects: fedora-all [bug 2265670]

Comment 3 Justin M. Forbes 2024-02-27 00:25:04 UTC

	Issue introduced in 5.9 with commit 638e4b825d52 and fixed in 6.1.75 with commit 62fca83303d6
	Issue introduced in 5.9 with commit 638e4b825d52 and fixed in 6.6.14 with commit f91cd728b10c
	Issue introduced in 5.9 with commit 638e4b825d52 and fixed in 6.7.2 with commit bfd9b20c4862
	Issue introduced in 5.9 with commit 638e4b825d52 and fixed in 6.8-rc1 with commit 876673364161

Comment 4 Justin M. Forbes 2024-02-27 00:25:37 UTC

This was fixed for Fedora with the 6.6.14 stable kernel updates.

Comment 6 Alex 2024-06-09 16:34:53 UTC

The result of automatic check (that is developed by Alexander Larkin) for this CVE-2023-52447 is: CHECK	Maybe valid. Check manually. with impact LOW (that is an approximation based on flags INIT BPF UAF  ; these flags parsed automatically based on patch data). Such automatic check happens only for Low/Moderates (and only when not from reporter, but parsing already existing CVE). Highs always checked manually (I check it myself and then we check it again in Remediation team). In rare cases some of the Moderates could be increased to High later.