Re: memcg/memcontrol02: fix fd leak in cleanup path
Disha Goel <[email protected]>
| Newsgroups | gmane.linux.ltp |
|---|---|
| Message-ID | <[email protected]> |
On 10/08/26 3:11 pm, Andrea Cervesato wrote: > Hi Disha, > >>> fd is initialized to 0 (a valid file descriptor) so the cleanup() >>> guard 'if (fd > -1)' would never trigger. >> >> Could this explanation be corrected? Zero is greater than -1, and the >> cleanup guard is added by this patch rather than being an existing guard. >> The relevant problem is that -1 is needed to distinguish an unopened >> descriptor before cleanup tries to close it. >> >>> + if (fd > -1) >>> + SAFE_CLOSE(fd); >> >> Could this use `fd != -1` instead? LTP file descriptors use -1 as the >> invalid sentinel, and the LTP convention requires this form for cleanup >> guards. > > this is correct. > >> >> --- [PATCH 2/2] --- >> >>> #include <inttypes.h> >>> >>> - tst_res(TINFO, "Child %d in %s: Allocating anon: %"PRIdPTR, >>> + tst_res(TINFO, "Child %d in %s: Allocating anon: %zu", >> >> Could the now-unused `<inttypes.h>` include be removed from both >> memcontrol03.c and memcontrol04.c? This patch replaces every `PRIdPTR` >> use in both files. > > And also this is correct. We should split the patch in two: one for > memcontrol03 and one for memcontrol04 > Hi Andrea, Thanks for the review. I've addressed all the comments and sent a v2. Link to v2: https://lore.kernel.org/ltp/20260818121436.1397-1-disgoel-tEXmvtCZX7AybS5Ee8rs3A@public.gmane.org/ > -- > Andrea Cervesato > SUSE QE Automation Engineer Linux > andrea.cervesato-IBi9RG/[email protected] -- Regards, Disha -- Mailing list info: https://lists.linux.it/listinfo/ltp