From: Yury Gribov <y.gribov@samsung.com>
To: Jakub Jelinek <jakub@redhat.com>
Cc: GCC Patches <gcc-patches@gcc.gnu.org>,
Andrey Belevantsev <abel@ispras.ru>,
Andrew MacLeod <amacleod@redhat.com>,
Andrew Pinski <pinskia@gmail.com>,
Diego Novillo <dnovillo@google.com>,
Geoff Keating <geoffk@geoffk.org>,
Jason Merrill <jason@redhat.com>,
Richard Biener <rguenther@suse.de>,
Steven Bosscher <steven@gcc.gnu.org>
Subject: Re: [PATCH 1/5] Fix asymmetric comparison functions
Date: Fri, 18 Dec 2015 19:40:00 -0000 [thread overview]
Message-ID: <56746138.9070709@samsung.com> (raw)
In-Reply-To: <5672A4D0.8090509@samsung.com>
On 12/17/2015 03:04 PM, Yury Gribov wrote:
> On 12/17/2015 02:39 PM, Jakub Jelinek wrote:
>> On Thu, Dec 17, 2015 at 11:58:30AM +0300, Yury Gribov wrote:
>>> 2015-12-17 Yury Gribov <tetra2005@gmail.com>
>>>
>>> * c-family/c-common.c (resort_field_decl_cmp):
>>> Make symmteric.
>>> * cp/class.c (method_name_cmp): Ditto.
>>> (resort_method_name_cmp): Ditto.
>>> * fortran/interface.c (pair_cmp): Ditto.
>>
>> Note, c-family, cp and fortran have their own ChangeLog files, so
>> the entries without those prefixes need to go into each one and can't
>> refer to other ChangeLog through Ditto/Likewise etc.
>> Typo in symmteric.
>
> Right, thanks.
>
>> That said, is this actually really a problem? I mean, is qsort
>> allowed to call the comparison function with the same arguments?
>> I think lots of the comparison functions just assume that
>> for int cmpfn (const void *x, const void *y) x != y.
>> And if qsort can't call the comparison function with the same argument,
>> then perhaps the caller has some knowledge your checker does not, say
>> that the entries that would compare equal by the comparison function
>> simply can't appear in the array (so the caller knows that the comparison
>> function should never return 0).
>
> Self-comparisons are certainly less dangerous than transitive ones. I
> personally not aware about libc's which can compare element to itself.
Jakub,
So it seems most people generally agree that self-comparisons (cmp(x,x)
== 0) are useless and don't need to be checked or fixed. What about
ensuring symmetry i.e. that cmp(x, y) == -cmp(y, x) forall x, y? One of
the bugs (pair_cmp in fortran/interface.c) is exactly about this.
> However
> * comparing an element to itself still a valid thing for qsort to do
> * most other comparison functions in GCC support this
>
>>> --- a/gcc/tree-vrp.c
>>> +++ b/gcc/tree-vrp.c
>>> @@ -5882,7 +5882,9 @@ compare_case_labels (const void *p1, const void
>>> *p2)
>>> else if (idx1 == idx2)
>>> {
>>> /* Make sure the default label is first in a group. */
>>> - if (!CASE_LOW (ci1->expr))
>>> + if (!CASE_LOW (ci1->expr) && !CASE_LOW (ci2->expr))
>>> + return 0;
>>> + else if (!CASE_LOW (ci1->expr))
>>> return -1;
>>> else if (!CASE_LOW (ci2->expr))
>>> return 1;
>>> --
>>> 1.9.1
>>
>> Say here, we know there is at most one default label in a switch, never
>> more. So, unless qsort is allowed to call compare_case_labels
>> with p1 == p2 (which really doesn't make sense), this case just won't
>> happen.
>
>
>
next prev parent reply other threads:[~2015-12-18 19:40 UTC|newest]
Thread overview: 33+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-12-17 8:55 [PATCH 0/5] Fix qsort " Yury Gribov
2015-12-17 8:58 ` [PATCH 1/5] Fix asymmetric " Yury Gribov
2015-12-17 11:39 ` Jakub Jelinek
2015-12-17 12:04 ` Yury Gribov
2015-12-18 19:40 ` Yury Gribov [this message]
2015-12-18 20:07 ` Jakub Jelinek
2015-12-18 20:09 ` Yury Gribov
2015-12-17 11:41 ` Richard Biener
2015-12-17 11:47 ` Yury Gribov
2015-12-17 11:57 ` Jakub Jelinek
2015-12-17 11:59 ` Richard Biener
2015-12-17 12:14 ` Yury Gribov
2015-12-17 12:25 ` Richard Biener
2015-12-17 12:36 ` Yury Gribov
2015-12-17 11:59 ` Andrey Belevantsev
2015-12-17 12:12 ` Yury Gribov
2015-12-17 12:48 ` Andrey Belevantsev
2015-12-17 19:39 ` Andrew Pinski
2015-12-17 22:00 ` Jason Merrill
2015-12-17 8:59 ` [PATCH 2/5] Fix more " Yury Gribov
2015-12-17 9:00 ` [PATCH 3/5] "Fix" intransitive comparison in reload_pseudo_compare_func Yury Gribov
2015-12-17 19:36 ` Vladimir Makarov
2015-12-18 19:50 ` Yury Gribov
2015-12-17 9:02 ` [PATCH 4/5] Fix intransitive comparison in compare_access_positions Yury Gribov
2015-12-17 14:58 ` Martin Jambor
2015-12-17 9:03 ` [PATCH 5/5] Fix intransitive comparison in dr_group_sort_cmp Yury Gribov
2015-12-17 11:57 ` Richard Biener
2015-12-17 12:33 ` Yury Gribov
2015-12-17 12:51 ` Richard Biener
2015-12-18 20:19 ` Yury Gribov
2015-12-18 22:30 ` Yuri Gribov
2015-12-25 11:42 ` Yury Gribov
2016-01-08 8:23 ` Richard Biener
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=56746138.9070709@samsung.com \
--to=y.gribov@samsung.com \
--cc=abel@ispras.ru \
--cc=amacleod@redhat.com \
--cc=dnovillo@google.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=geoffk@geoffk.org \
--cc=jakub@redhat.com \
--cc=jason@redhat.com \
--cc=pinskia@gmail.com \
--cc=rguenther@suse.de \
--cc=steven@gcc.gnu.org \
/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).