On Thu, Jun 1, 2023 at 11:26 AM Yosry Ahmed <yosryahmed@xxxxxxxxxx> wrote:
>
> On Thu, Jun 1, 2023 at 11:18 AM Nathan Chancellor <nathan@xxxxxxxxxx> wrote:
> >
> > On Thu, Jun 01, 2023 at 11:09:48AM -0700, Yosry Ahmed wrote:
> > > On Thu, Jun 1, 2023 at 8:26 AM Nathan Chancellor <nathan@xxxxxxxxxx> wrote:
> > > >
> > > > Clang warns:
> > > >
> > > > mm/zswap.c:1183:6: error: variable 'ret' is used uninitialized whenever 'if' condition is true [-Werror,-Wsometimes-uninitialized]
> > > > 1183 | if (objcg && !obj_cgroup_may_zswap(objcg))
> > > > | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> > > > mm/zswap.c:1327:9: note: uninitialized use occurs here
> > > > 1327 | return ret;
> > > > | ^~~
> > > > mm/zswap.c:1183:2: note: remove the 'if' if its condition is always false
> > > > 1183 | if (objcg && !obj_cgroup_may_zswap(objcg))
> > > > | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> > > > 1184 | goto reject;
> > > > | ~~~~~~~~~~~~~~~~~~~~~~~~~~~
> > > > mm/zswap.c:1158:9: note: initialize the variable 'ret' to silence this warning
> > > > 1158 | int ret;
> > > > | ^
> > > > | = 0
> > > > 1 error generated.
Ooops that's my bad. Thanks for the fix!
> > > >
> > > > Set ret to -EINVAL, which seems to best match the situation the comment
> > > > mentions and matches an earlier bailout due to an unsupported type or
> > > > situation.
> > >
> > > Shouldn't this be -ENOMEM? I think this was the error code returned in
> > > this situation before 6804144bf1cf.
> >
> > Sure, I can update this to be -ENOMEM if that is appropriate. I am not
> > sure it actually matters though, since as far as I can tell, this
> > function is only called via the ->store() call in __frontswap_store(),
> > which is only called by frontswap_store(), which in turn is only called by
> > swap_writepage() and its return value is only checked against 0, no
> > other error values. I could be missing a macro call though.
> >
> > If I should send a v2, please let me know.
>
> You are right that no one actually checks the error code now, but it
> would be nice to stay consistent, just in case we want to do proper
> error handling at some point :)
I agree with Yosry - I think ENOMEM is more appropriate here as well.
>
> >
> > Cheers,
> > Nathan
> >
> > > > Fixes: 6804144bf1cf ("zswap: do not shrink if cgroup may not zswap")
> > > > Reported-by: kernel test robot <lkp@xxxxxxxxx>
> > > > Closes: https://lore.kernel.org/202306011435.2BxsHFUE-lkp@xxxxxxxxx/
> > > > Closes: https://lore.kernel.org/202306012152.o4FL7Y6J-lkp@xxxxxxxxx/
> > > > Closes: https://github.com/ClangBuiltLinux/linux/issues/1861
> > > > Signed-off-by: Nathan Chancellor <nathan@xxxxxxxxxx>
> > > > ---
> > > > mm/zswap.c | 4 +++-
> > > > 1 file changed, 3 insertions(+), 1 deletion(-)
> > > >
> > > > diff --git a/mm/zswap.c b/mm/zswap.c
> > > > index cff93643a6ab..e8313e2f5807 100644
> > > > --- a/mm/zswap.c
> > > > +++ b/mm/zswap.c
> > > > @@ -1180,8 +1180,10 @@ static int zswap_frontswap_store(unsigned type, pgoff_t offset,
> > > > * local cgroup limits.
> > > > */
> > > > objcg = get_obj_cgroup_from_page(page);
> > > > - if (objcg && !obj_cgroup_may_zswap(objcg))
> > > > + if (objcg && !obj_cgroup_may_zswap(objcg)) {
> > > > + ret = -EINVAL;
> > > > goto reject;
> > > > + }
> > > >
> > > > /* reclaim space if needed */
> > > > if (zswap_is_full()) {
> > > >
> > > > ---
> > > > base-commit: b2424568bd9b947b2698b693ff24fec63bb4b36a
> > > > change-id: 20230601-zswap-cgroup-wsometimes-uninitialized-b549ff4247ed
> > > >
> > > > Best regards,
> > > > --
> > > > Nathan Chancellor <nathan@xxxxxxxxxx>
> > > >
> > > >
