diff options
author | Viacheslav Dubeyko <Slava.Dubeyko@ibm.com> | 2025-09-02 12:08:45 -0700 |
---|---|---|
committer | Ilya Dryomov <idryomov@gmail.com> | 2025-10-08 23:30:47 +0200 |
commit | c66120c84295a0495eb46dcfba829457acd6ef7d (patch) | |
tree | 0e99f320515747ef99de76b3e5abec9f2a059c59 /rust/helpers/pci.c | |
parent | 98a2850de49c10a1a09642e17978b925f95e6029 (diff) |
ceph: cleanup in ceph_alloc_readdir_reply_buffer()
The Coverity Scan service has reported potential issue
in ceph_alloc_readdir_reply_buffer() [1]. If order could
be negative one, then it expects the issue in the logic:
num_entries = (PAGE_SIZE << order) / size;
Technically speaking, this logic [2] should prevent from
making the order variable negative:
if (!rinfo->dir_entries)
return -ENOMEM;
However, the allocation logic requires some cleanup.
This patch makes sure that calculated bytes count
will never exceed ULONG_MAX before get_order()
calculation. And it adds the checking of order
variable on negative value to guarantee that second
half of the function's code will never operate by
negative value of order variable even if something
will be wrong or to be changed in the first half of
the function's logic.
v2
Alex Markuze suggested to add unlikely() macro
for introduced condition checks.
[1] https://scan5.scan.coverity.com/#/project-view/64304/10063?selectedIssue=1198252
[2] https://elixir.bootlin.com/linux/v6.17-rc3/source/fs/ceph/mds_client.c#L2553
Signed-off-by: Viacheslav Dubeyko <Slava.Dubeyko@ibm.com>
Reviewed-by: Alex Markuze <amarkuze@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Diffstat (limited to 'rust/helpers/pci.c')
0 files changed, 0 insertions, 0 deletions