* Re: [PATCH] tree-optimization/107561 - reduce -Wstringop-overflow false positives
[not found] <20230329121203.002E43858434@sourceware.org>
@ 2023-03-31 14:13 ` Jeff Law
0 siblings, 0 replies; 3+ messages in thread
From: Jeff Law @ 2023-03-31 14:13 UTC (permalink / raw)
To: gcc-patches
On 3/29/23 06:11, Richard Biener via Gcc-patches wrote:
> The following tells pointer-query to prefer a zero size when we
> are querying for the size range for a write into an object we've
> determined is of zero size. That avoids diagnostics about really
> varying size arguments that just get a meaningful range for example
> because they are multiplied by an element size.
>
> I've adjusted only one call to get_size_range since that's what
> I have a testcase for. I think this is the most sensible
> "workaround" for some of the false positives we see.
>
> Bootstrapped on x86_64-unknown-linux-gnu, testing in progress.
>
> OK if it succeeds?
>
> Thanks,
> Richard.
>
> PR tree-optimization/107561
> * gimple-ssa-warn-access.cc (get_size_range): Add flags
> argument and pass it on.
> (check_access): When querying for the size range pass
> SR_ALLOW_ZERO when the known destination size is zero.
>
> * g++.dg/pr71488.C: Remove XFAILed bogus diagnostic again.
> * g++.dg/warn/Warray-bounds-16.C: Likewise.
OK.
jeff
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] tree-optimization/107561 - reduce -Wstringop-overflow false positives
@ 2023-03-30 11:16 Richard Biener
0 siblings, 0 replies; 3+ messages in thread
From: Richard Biener @ 2023-03-30 11:16 UTC (permalink / raw)
To: gcc-patches; +Cc: Jakub Jelinek
On Wed, 29 Mar 2023, Richard Biener wrote:
> The following tells pointer-query to prefer a zero size when we
> are querying for the size range for a write into an object we've
> determined is of zero size. That avoids diagnostics about really
> varying size arguments that just get a meaningful range for example
> because they are multiplied by an element size.
>
> I've adjusted only one call to get_size_range since that's what
> I have a testcase for. I think this is the most sensible
> "workaround" for some of the false positives we see.
>
> Bootstrapped on x86_64-unknown-linux-gnu, testing in progress.
>
> OK if it succeeds?
It did, I have now pushed this, it seems like an obvious improvement.
Richard.
> Thanks,
> Richard.
>
> PR tree-optimization/107561
> * gimple-ssa-warn-access.cc (get_size_range): Add flags
> argument and pass it on.
> (check_access): When querying for the size range pass
> SR_ALLOW_ZERO when the known destination size is zero.
>
> * g++.dg/pr71488.C: Remove XFAILed bogus diagnostic again.
> * g++.dg/warn/Warray-bounds-16.C: Likewise.
> ---
> gcc/gimple-ssa-warn-access.cc | 19 +++++++++++++------
> gcc/testsuite/g++.dg/pr71488.C | 1 -
> gcc/testsuite/g++.dg/warn/Warray-bounds-16.C | 1 -
> 3 files changed, 13 insertions(+), 8 deletions(-)
>
> diff --git a/gcc/gimple-ssa-warn-access.cc b/gcc/gimple-ssa-warn-access.cc
> index 88d44690ade..b3de4b77924 100644
> --- a/gcc/gimple-ssa-warn-access.cc
> +++ b/gcc/gimple-ssa-warn-access.cc
> @@ -1198,10 +1198,10 @@ warn_for_access (location_t loc, tree func, tree expr, int opt,
>
> static void
> get_size_range (range_query *query, tree bound, gimple *stmt, tree range[2],
> - const offset_int bndrng[2])
> + int flags, const offset_int bndrng[2])
> {
> if (bound)
> - get_size_range (query, bound, stmt, range);
> + get_size_range (query, bound, stmt, range, flags);
>
> if (!bndrng || (bndrng[0] == 0 && bndrng[1] == HOST_WIDE_INT_M1U))
> return;
> @@ -1347,7 +1347,12 @@ check_access (GimpleOrTree exp, tree dstwrite,
> /* Set RANGE to that of DSTWRITE if non-null, bounded by PAD->DST_BNDRNG
> if valid. */
> gimple *stmt = pad ? pad->stmt : nullptr;
> - get_size_range (rvals, dstwrite, stmt, range, pad ? pad->dst_bndrng : NULL);
> + get_size_range (rvals, dstwrite, stmt, range,
> + /* If the destination has known zero size prefer a zero
> + size range to avoid false positives if that's a
> + possibility. */
> + integer_zerop (dstsize) ? SR_ALLOW_ZERO : 0,
> + pad ? pad->dst_bndrng : NULL);
>
> tree func = get_callee_fndecl (exp);
> /* Read vs write access by built-ins can be determined from the const
> @@ -1442,7 +1447,8 @@ check_access (GimpleOrTree exp, tree dstwrite,
> {
> /* Set RANGE to that of MAXREAD, bounded by PAD->SRC_BNDRNG if
> PAD is nonnull and BNDRNG is valid. */
> - get_size_range (rvals, maxread, stmt, range, pad ? pad->src_bndrng : NULL);
> + get_size_range (rvals, maxread, stmt, range, 0,
> + pad ? pad->src_bndrng : NULL);
>
> location_t loc = get_location (exp);
> tree size = dstsize;
> @@ -1489,7 +1495,8 @@ check_access (GimpleOrTree exp, tree dstwrite,
> {
> /* Set RANGE to that of MAXREAD, bounded by PAD->SRC_BNDRNG if
> PAD is nonnull and BNDRNG is valid. */
> - get_size_range (rvals, maxread, stmt, range, pad ? pad->src_bndrng : NULL);
> + get_size_range (rvals, maxread, stmt, range, 0,
> + pad ? pad->src_bndrng : NULL);
> /* Set OVERREAD for reads starting just past the end of an object. */
> overread = pad->src.sizrng[1] - pad->src.offrng[0] < pad->src_bndrng[0];
> range[0] = wide_int_to_tree (sizetype, pad->src_bndrng[0]);
> @@ -2679,7 +2686,7 @@ pass_waccess::check_strncmp (gcall *stmt)
> /* Determine the range of the bound first and bail if it fails; it's
> cheaper than computing the size of the objects. */
> tree bndrng[2] = { NULL_TREE, NULL_TREE };
> - get_size_range (m_ptr_qry.rvals, bound, stmt, bndrng, adata1.src_bndrng);
> + get_size_range (m_ptr_qry.rvals, bound, stmt, bndrng, 0, adata1.src_bndrng);
> if (!bndrng[0] || integer_zerop (bndrng[0]))
> return;
>
> diff --git a/gcc/testsuite/g++.dg/pr71488.C b/gcc/testsuite/g++.dg/pr71488.C
> index a89e6465a94..ffe30f1afeb 100644
> --- a/gcc/testsuite/g++.dg/pr71488.C
> +++ b/gcc/testsuite/g++.dg/pr71488.C
> @@ -3,7 +3,6 @@
> // { dg-options "-O3 -std=c++11" }
> // { dg-additional-options "-msse4" { target sse4_runtime } }
> // { dg-require-effective-target c++11 }
> -// { dg-bogus "size 0 overflows" "pr107561" { xfail *-*-* } 0 }
>
> #include <valarray>
>
> diff --git a/gcc/testsuite/g++.dg/warn/Warray-bounds-16.C b/gcc/testsuite/g++.dg/warn/Warray-bounds-16.C
> index 49d3437fa1b..89cbadb91c7 100644
> --- a/gcc/testsuite/g++.dg/warn/Warray-bounds-16.C
> +++ b/gcc/testsuite/g++.dg/warn/Warray-bounds-16.C
> @@ -20,7 +20,6 @@ struct S
>
> for (int i = 0; i < m; i++)
> new (p + i) int (); /* { dg-bogus "bounds" "pr102690" { xfail *-*-* } } */
> - // { dg-bogus "size 0 overflows" "pr107561" { xfail ilp32 } .-1 }
> }
> };
>
>
--
Richard Biener <rguenther@suse.de>
SUSE Software Solutions Germany GmbH, Frankenstrasse 146, 90461 Nuernberg,
Germany; GF: Ivo Totev, Andrew Myers, Andrew McDonald, Boudien Moerman;
HRB 36809 (AG Nuernberg)
^ permalink raw reply [flat|nested] 3+ messages in thread
* [PATCH] tree-optimization/107561 - reduce -Wstringop-overflow false positives
@ 2023-03-29 12:11 Richard Biener
0 siblings, 0 replies; 3+ messages in thread
From: Richard Biener @ 2023-03-29 12:11 UTC (permalink / raw)
To: gcc-patches; +Cc: Jakub Jelinek
The following tells pointer-query to prefer a zero size when we
are querying for the size range for a write into an object we've
determined is of zero size. That avoids diagnostics about really
varying size arguments that just get a meaningful range for example
because they are multiplied by an element size.
I've adjusted only one call to get_size_range since that's what
I have a testcase for. I think this is the most sensible
"workaround" for some of the false positives we see.
Bootstrapped on x86_64-unknown-linux-gnu, testing in progress.
OK if it succeeds?
Thanks,
Richard.
PR tree-optimization/107561
* gimple-ssa-warn-access.cc (get_size_range): Add flags
argument and pass it on.
(check_access): When querying for the size range pass
SR_ALLOW_ZERO when the known destination size is zero.
* g++.dg/pr71488.C: Remove XFAILed bogus diagnostic again.
* g++.dg/warn/Warray-bounds-16.C: Likewise.
---
gcc/gimple-ssa-warn-access.cc | 19 +++++++++++++------
gcc/testsuite/g++.dg/pr71488.C | 1 -
gcc/testsuite/g++.dg/warn/Warray-bounds-16.C | 1 -
3 files changed, 13 insertions(+), 8 deletions(-)
diff --git a/gcc/gimple-ssa-warn-access.cc b/gcc/gimple-ssa-warn-access.cc
index 88d44690ade..b3de4b77924 100644
--- a/gcc/gimple-ssa-warn-access.cc
+++ b/gcc/gimple-ssa-warn-access.cc
@@ -1198,10 +1198,10 @@ warn_for_access (location_t loc, tree func, tree expr, int opt,
static void
get_size_range (range_query *query, tree bound, gimple *stmt, tree range[2],
- const offset_int bndrng[2])
+ int flags, const offset_int bndrng[2])
{
if (bound)
- get_size_range (query, bound, stmt, range);
+ get_size_range (query, bound, stmt, range, flags);
if (!bndrng || (bndrng[0] == 0 && bndrng[1] == HOST_WIDE_INT_M1U))
return;
@@ -1347,7 +1347,12 @@ check_access (GimpleOrTree exp, tree dstwrite,
/* Set RANGE to that of DSTWRITE if non-null, bounded by PAD->DST_BNDRNG
if valid. */
gimple *stmt = pad ? pad->stmt : nullptr;
- get_size_range (rvals, dstwrite, stmt, range, pad ? pad->dst_bndrng : NULL);
+ get_size_range (rvals, dstwrite, stmt, range,
+ /* If the destination has known zero size prefer a zero
+ size range to avoid false positives if that's a
+ possibility. */
+ integer_zerop (dstsize) ? SR_ALLOW_ZERO : 0,
+ pad ? pad->dst_bndrng : NULL);
tree func = get_callee_fndecl (exp);
/* Read vs write access by built-ins can be determined from the const
@@ -1442,7 +1447,8 @@ check_access (GimpleOrTree exp, tree dstwrite,
{
/* Set RANGE to that of MAXREAD, bounded by PAD->SRC_BNDRNG if
PAD is nonnull and BNDRNG is valid. */
- get_size_range (rvals, maxread, stmt, range, pad ? pad->src_bndrng : NULL);
+ get_size_range (rvals, maxread, stmt, range, 0,
+ pad ? pad->src_bndrng : NULL);
location_t loc = get_location (exp);
tree size = dstsize;
@@ -1489,7 +1495,8 @@ check_access (GimpleOrTree exp, tree dstwrite,
{
/* Set RANGE to that of MAXREAD, bounded by PAD->SRC_BNDRNG if
PAD is nonnull and BNDRNG is valid. */
- get_size_range (rvals, maxread, stmt, range, pad ? pad->src_bndrng : NULL);
+ get_size_range (rvals, maxread, stmt, range, 0,
+ pad ? pad->src_bndrng : NULL);
/* Set OVERREAD for reads starting just past the end of an object. */
overread = pad->src.sizrng[1] - pad->src.offrng[0] < pad->src_bndrng[0];
range[0] = wide_int_to_tree (sizetype, pad->src_bndrng[0]);
@@ -2679,7 +2686,7 @@ pass_waccess::check_strncmp (gcall *stmt)
/* Determine the range of the bound first and bail if it fails; it's
cheaper than computing the size of the objects. */
tree bndrng[2] = { NULL_TREE, NULL_TREE };
- get_size_range (m_ptr_qry.rvals, bound, stmt, bndrng, adata1.src_bndrng);
+ get_size_range (m_ptr_qry.rvals, bound, stmt, bndrng, 0, adata1.src_bndrng);
if (!bndrng[0] || integer_zerop (bndrng[0]))
return;
diff --git a/gcc/testsuite/g++.dg/pr71488.C b/gcc/testsuite/g++.dg/pr71488.C
index a89e6465a94..ffe30f1afeb 100644
--- a/gcc/testsuite/g++.dg/pr71488.C
+++ b/gcc/testsuite/g++.dg/pr71488.C
@@ -3,7 +3,6 @@
// { dg-options "-O3 -std=c++11" }
// { dg-additional-options "-msse4" { target sse4_runtime } }
// { dg-require-effective-target c++11 }
-// { dg-bogus "size 0 overflows" "pr107561" { xfail *-*-* } 0 }
#include <valarray>
diff --git a/gcc/testsuite/g++.dg/warn/Warray-bounds-16.C b/gcc/testsuite/g++.dg/warn/Warray-bounds-16.C
index 49d3437fa1b..89cbadb91c7 100644
--- a/gcc/testsuite/g++.dg/warn/Warray-bounds-16.C
+++ b/gcc/testsuite/g++.dg/warn/Warray-bounds-16.C
@@ -20,7 +20,6 @@ struct S
for (int i = 0; i < m; i++)
new (p + i) int (); /* { dg-bogus "bounds" "pr102690" { xfail *-*-* } } */
- // { dg-bogus "size 0 overflows" "pr107561" { xfail ilp32 } .-1 }
}
};
--
2.35.3
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2023-03-31 14:13 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
[not found] <20230329121203.002E43858434@sourceware.org>
2023-03-31 14:13 ` [PATCH] tree-optimization/107561 - reduce -Wstringop-overflow false positives Jeff Law
2023-03-30 11:16 Richard Biener
-- strict thread matches above, loose matches on Subject: below --
2023-03-29 12:11 Richard Biener
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).