Bug 2363352 (CVE-2025-37753) - CVE-2025-37753 kernel: s390/cpumf: Fix double free on error in cpumf_pmu_event_init()
Summary: CVE-2025-37753 kernel: s390/cpumf: Fix double free on error in cpumf_pmu_even...
Keywords:
Status: NEW
Alias: CVE-2025-37753
Product: Security Response
Classification: Other
Component: vulnerability
Version: unspecified
Hardware: All
OS: Linux
unspecified
unspecified
Target Milestone: ---
Assignee: Product Security DevOps Team
QA Contact:
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2025-05-01 14:04 UTC by OSIDB Bzimport
Modified: 2025-06-18 10:31 UTC (History)
4 users (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed:
Embargoed:


Attachments (Terms of Use)

Description OSIDB Bzimport 2025-05-01 14:04:55 UTC
In the Linux kernel, the following vulnerability has been resolved:

s390/cpumf: Fix double free on error in cpumf_pmu_event_init()

In PMU event initialization functions
 - cpumsf_pmu_event_init()
 - cpumf_pmu_event_init()
 - cfdiag_event_init()
the partially created event had to be removed when an error was detected.
The event::event_init() member function had to release all resources
it allocated in case of error. event::destroy() had to be called
on freeing an event after it was successfully created and
event::event_init() returned success.

With

commit c70ca298036c ("perf/core: Simplify the perf_event_alloc() error path")

this is not necessary anymore. The performance subsystem common
code now always calls event::destroy() to clean up the allocated
resources created during event initialization.

Remove the event::destroy() invocation in PMU event initialization
or that function is called twice for each event that runs into an
error condition in event creation.

This is the kernel log entry which shows up without the fix:

------------[ cut here ]------------
refcount_t: underflow; use-after-free.
WARNING: CPU: 0 PID: 43388 at lib/refcount.c:87	refcount_dec_not_one+0x74/0x90
CPU: 0 UID: 0 PID: 43388 Comm: perf Not tainted 6.15.0-20250407.rc1.git0.300.fc41.s390x+git #1 NONE
Hardware name: IBM 3931 A01 704 (LPAR)
Krnl PSW : 0704c00180000000 00000209cb2c1b88 (refcount_dec_not_one+0x78/0x90)
           R:0 T:1 IO:1 EX:1 Key:0 M:1 W:0 P:0 AS:3 CC:0 PM:0 RI:0 EA:3
Krnl GPRS: 0000020900000027 0000020900000023 0000000000000026 0000018900000000
           00000004a2200a00 0000000000000000 0000000000000057 ffffffffffffffea
           00000002b386c600 00000002b3f5b3e0 00000209cc51f140 00000209cc7fc550
           0000000001449d38 ffffffffffffffff 00000209cb2c1b84 00000189d67dfb80
Krnl Code: 00000209cb2c1b78: c02000506727	larl	%r2,00000209cbcce9c6
           00000209cb2c1b7e: c0e5ffbd4431	brasl	%r14,00000209caa6a3e0
          #00000209cb2c1b84: af000000		mc	0,0
          >00000209cb2c1b88: a7480001		lhi	%r4,1
           00000209cb2c1b8c: ebeff0a00004	lmg	%r14,%r15,160(%r15)
           00000209cb2c1b92: ec243fbf0055	risbg	%r2,%r4,63,191,0
           00000209cb2c1b98: 07fe		bcr	15,%r14
           00000209cb2c1b9a: 47000700		bc	0,1792
Call Trace:
 [<00000209cb2c1b88>] refcount_dec_not_one+0x78/0x90
 [<00000209cb2c1dc4>] refcount_dec_and_mutex_lock+0x24/0x90
 [<00000209caa3c29e>] hw_perf_event_destroy+0x2e/0x80
 [<00000209cacaf8b4>] __free_event+0x74/0x270
 [<00000209cacb47c4>] perf_event_alloc.part.0+0x4a4/0x730
 [<00000209cacbf3e8>] __do_sys_perf_event_open+0x248/0xc20
 [<00000209cacc14a4>] __s390x_sys_perf_event_open+0x44/0x50
 [<00000209cb8114de>] __do_syscall+0x12e/0x260
 [<00000209cb81ce34>] system_call+0x74/0x98
Last Breaking-Event-Address:
 [<00000209caa6a4d2>] __warn_printk+0xf2/0x100
---[ end trace 0000000000000000 ]---

Comment 1 Avinash Hanwate 2025-05-02 05:10:46 UTC
Upstream advisory:
https://lore.kernel.org/linux-cve-announce/2025050138-CVE-2025-37753-9fd7@gregkh/T

Comment 2 TEJ RATHI 2025-05-29 04:06:57 UTC
This CVE has been rejected by the Linux kernel community. Refer to the announcement: https://lore.kernel.org/linux-cve-announce/2025052625-REJECTED-1365@gregkh/

Comment added by: Automated Script


Note You need to log in before you can comment on or make changes to this bug.