From: Jeff Law <law@redhat.com>
To: Martin Sebor <msebor@gmail.com>, Jakub Jelinek <jakub@redhat.com>
Cc: Gcc Patch List <gcc-patches@gcc.gnu.org>
Subject: Re: [PATCH] make -Wrestrict for strcat more meaningful (PR 83698)
Date: Wed, 14 Feb 2018 05:36:00 -0000 [thread overview]
Message-ID: <e3bf9918-c4a8-3bac-d16e-d6e98b9f06ab@redhat.com> (raw)
In-Reply-To: <3fe66f05-1c06-ada0-b27b-b4a017677851@gmail.com>
On 01/16/2018 05:35 PM, Martin Sebor wrote:
> On 01/16/2018 02:32 PM, Jakub Jelinek wrote:
>> On Tue, Jan 16, 2018 at 01:36:26PM -0700, Martin Sebor wrote:
>>> --- gcc/gimple-ssa-warn-restrict.c   (revision 256752)
>>> +++ gcc/gimple-ssa-warn-restrict.c   (working copy)
>>> @@ -384,6 +384,12 @@ builtin_memref::builtin_memref (tree expr, tree si
>>> Â Â Â Â Â Â base = SSA_NAME_VAR (base);
>>> Â Â Â Â Â Â }
>>>
>>> +Â if (DECL_P (base) && TREE_CODE (TREE_TYPE (base)) == ARRAY_TYPE)
>>> +Â Â Â {
>>> +Â Â Â Â Â if (offrange[0] < 0 && offrange[1] > 0)
>>> +Â Â Â offrange[0] = 0;
>>> +Â Â Â }
>>
>> Why the 2 nested ifs?
>
> No particular reason. There may have been more code in there
> that I ended up removing. Or a comment. I can remove the
> extra braces when the patch is approved.
>
>>
>>> @@ -1079,14 +1085,35 @@ builtin_access::strcat_overlap ()
>>> Â Â Â Â return false;
>>>
>>> Â Â /* When strcat overlap is certain it is always a single byte:
>>> -    the terminatinn NUL, regardless of offsets and sizes. When
>>> +    the terminating NUL, regardless of offsets and sizes. When
>>>      overlap is only possible its range is [0, 1]. */
>>> Â Â acs.ovlsiz[0] = dstref->sizrange[0] == dstref->sizrange[1] ? 1 : 0;
>>> Â Â acs.ovlsiz[1] = 1;
>>> -Â acs.ovloff[0] = (dstref->sizrange[0] +
>>> dstref->offrange[0]).to_shwi ();
>>> -Â acs.ovloff[1] = (dstref->sizrange[1] +
>>> dstref->offrange[1]).to_shwi ();
>>
>> You use to_shwi many times in the patch, do the callers or something
>> earlier
>> in this function guarantee that you aren't throwing away any bits (unlike
>> tree_to_shwi, to_shwi method doesn't ICE, just throws away upper bits).
>> Especially when you perform additions like here, even if both
>> wide_ints fit
>> into a shwi, the result might not.
>
> No, I'm not sure. In fact, it wouldn't surprise me if it did
> happen. It doesn't cause false positives or negatives but it
> can make the offsets less than meaningful in cases where they
> are within valid bounds. There are also cases where they are
> meaningless to begin with and there is little the pass can do
> about that.
I was kind of expecting an update to try and address some of these
issues. Though after re-reading your response the consequence of
throwing away bits here is just the diagnostic is not as precise as it
could be, right? ie, it doesn't change when we issue a diagnostic, just
the contents of the diagnostic.
I filed this into my gcc9 bucket because it doesn't fix a regression,
but it appears that a regression fix does depend on this stuff to some
degree (84095). So I'll try to take a look at this shortly so that we
can unblock 84095.
Jeff
next prev parent reply other threads:[~2018-02-14 5:36 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-01-16 20:41 Martin Sebor
2018-01-16 21:36 ` Jakub Jelinek
2018-01-17 0:54 ` Martin Sebor
2018-01-30 18:29 ` Martin Sebor
2018-02-06 3:20 ` [PING #2] " Martin Sebor
2018-02-13 3:15 ` [PING #3] " Martin Sebor
2018-02-14 5:36 ` Jeff Law [this message]
2018-02-14 5:54 ` Jeff Law
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=e3bf9918-c4a8-3bac-d16e-d6e98b9f06ab@redhat.com \
--to=law@redhat.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=jakub@redhat.com \
--cc=msebor@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).