URL:
https://github.com/SSSD/sssd/pull/140
Title: #140: gcc7 related fixes
lslebodn commented:
"""
On (31/01/17 04:06), fidencio wrote:
On Tue, Jan 31, 2017 at 10:29 AM, lslebodn
<notifications(a)github.com> wrote:
>
> On (30/01/17 23:52), fidencio wrote:
> >fidencio requested changes on this pull request.
> >
> >Neither the short log or the log are consistent between the patches.
> >Please, let's try to follow the git-commit-template.
> >
> >Also, there are some questions in two of the patches, which are basically
curiosity.
> >
>
> Could you be more specific?
> There is just missing ticket in messages. Because ther is not a ticket.
* sssctl: Fix warning may be used uninitialized:
Component should be in capital letters according to the git-commit-template
git log says something different
sh$ git log --oneline | grep " SSSCTL:" | wc -l
1
sh$ git log --oneline | grep " sssctl:" | wc -l
26
And I cannot see a reason to change it.
sssctl looks better in lowercase. becaus eutility is sssctl and not SSSCTL
* pam_sss: Suppress warning format-truncation
Same comment
the same here.
* Suppres implicit-fallthrough from gcc 7
Some kind of comments are recognized by gcc 7
but they are ignored with -Wimplicit-fallthrough=5
and only attributes disable the warning.
Please, don't break the lines with 52 characters. Do it on 72 as recommended.
I know the recommendation is "do not exceed 72 (...)" and you don't
exceed. But we're not going to take this path as someone may write a
commit message with a word per-line just because it doesn't exceed the
72 characters :-)
If it is the only problem then I can change it before pushing patches :-)
>
>
> >> @@ -434,8 +434,8 @@ static errno_t sssctl_fetch_object(TALLOC_CTX *mem_ctx,
> > struct sss_domain_info **_dom)
> > {
> > TALLOC_CTX *tmp_ctx;
> >- struct sysdb_attrs *entry;
> >- struct sss_domain_info *dom;
> >+ struct sysdb_attrs *entry = NULL;
> >+ struct sss_domain_info *dom = NULL;
> >
> >I'm curious why just this code path triggered the issue.
> >We have similar code in a lot of other places (even in the same file).
> >
>
> Could you be more specific?
> Because sssctl_find_object is called just once and there is something
> smelly (hyper optimized) there.
Let me try to re-phrase. Is there any other part of the code where we
can be bite by the very same issue?
In a lot of places we don't initialize the output vars and may be the
case we end up hitting this very same "smelly hyper optimization" as
well.
I cannot see any other warning related to uninitialized data.
Does it answer your question?
LS
"""
See the full comment at
https://github.com/SSSD/sssd/pull/140#issuecomment-276361908