public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug c++/105994] New: Incorrect optimization with -ftree-sra and casting (strictly compatible) refs
@ 2022-06-16 0:10 gcc_bugzilla.calin79 at neverbox dot com
2022-06-16 0:21 ` [Bug c++/105994] " pinskia at gcc dot gnu.org
` (4 more replies)
0 siblings, 5 replies; 6+ messages in thread
From: gcc_bugzilla.calin79 at neverbox dot com @ 2022-06-16 0:10 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105994
Bug ID: 105994
Summary: Incorrect optimization with -ftree-sra and casting
(strictly compatible) refs
Product: gcc
Version: 13.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c++
Assignee: unassigned at gcc dot gnu.org
Reporter: gcc_bugzilla.calin79 at neverbox dot com
Target Milestone: ---
This code (also in attachment) should return 106, but instead returns 14:
////////////////////////////////////////////////////////////////
struct bar
{
int id;
};
struct foo
{
int id;
};
inline const bar& h(const foo& t)
{
return reinterpret_cast<const bar&>(t);
}
int g(foo t)
{
const bar& test = h(foo{t});
if(test.id == 5)
return 7;
return test.id;
}
int main()
{
int a = g(foo{5});
int b = g(foo{99});
return a + b;
}
////////////////////////////////////////////////////////////////
It happens with -O1 - and more specifically, as long as `-finline -ftree-ccp
-ftree-sra` are enabled.
Version: gcc version 13.0.0 20220615 (experimental) (GCC)
built from latest master as of 2022-06-15,
dc9b92facf87a6f2d8b0e5d5fc404f30c3b15a74
Target: x86_64-pc-linux-gnu
Configured with: ../src/configure
--prefix=/opt/compiler-explorer/gcc-build/install --enable-languages=c++
Triggered with:
g++ -std=c++20 -O1 -o buggy_bin_14 gcc_bug_reference_cast.cpp
# should print 106 (7 + 99)
./buggy_bin_14; echo $?
# instead prints 14 - g() always returns 7
Shortest number of optimizations that triggers it:
g++ -std=c++20 -O1 -fno-auto-inc-dec -fno-branch-count-reg
-fno-combine-stack-adjustments -fno-compare-elim -fno-cprop-registers -fno-dce
-fno-defer-pop -fno-delayed-branch -fno-dse -fno-forward-propagate
-fno-guess-branch-probability -fno-if-conversion -fno-if-conversion2
-fno-inline-functions-called-once -fno-ipa-profile -fno-ipa-pure-const
-fno-ipa-reference -fno-ipa-reference-addressable -fno-merge-constants
-fno-move-loop-invariants -fno-omit-frame-pointer -fno-reorder-blocks
-fno-shrink-wrap -fno-shrink-wrap-separate -fno-split-wide-types
-fno-ssa-backprop -fno-ssa-phiopt -fno-tree-bit-ccp -fno-tree-ch
-fno-tree-coalesce-vars -fno-tree-copy-prop -fno-tree-dce
-fno-tree-dominator-opts -fno-tree-dse -fno-tree-forwprop -fno-tree-fre
-fno-tree-phiprop -fno-tree-pta -fno-tree-scev-cprop -fno-tree-sink
-fno-tree-slsr -fno-tree-ter -fno-unit-at-a-time -finline -ftree-ccp -ftree-sra
-o buggy_bin_14 gcc_bug_reference_cast.cpp
Compiler output:
None
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug c++/105994] Incorrect optimization with -ftree-sra and casting (strictly compatible) refs
2022-06-16 0:10 [Bug c++/105994] New: Incorrect optimization with -ftree-sra and casting (strictly compatible) refs gcc_bugzilla.calin79 at neverbox dot com
@ 2022-06-16 0:21 ` pinskia at gcc dot gnu.org
2022-06-16 0:22 ` pinskia at gcc dot gnu.org
` (3 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: pinskia at gcc dot gnu.org @ 2022-06-16 0:21 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105994
--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Since bar and foo are not the same/compatible types this violates aliasing
rules. They might be layout compatible but they are compatible. Two different
things.
Also
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug c++/105994] Incorrect optimization with -ftree-sra and casting (strictly compatible) refs
2022-06-16 0:10 [Bug c++/105994] New: Incorrect optimization with -ftree-sra and casting (strictly compatible) refs gcc_bugzilla.calin79 at neverbox dot com
2022-06-16 0:21 ` [Bug c++/105994] " pinskia at gcc dot gnu.org
@ 2022-06-16 0:22 ` pinskia at gcc dot gnu.org
2022-06-16 0:23 ` pinskia at gcc dot gnu.org
` (2 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: pinskia at gcc dot gnu.org @ 2022-06-16 0:22 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105994
--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Hmm, but at -O1, strict aliasing is off so this should work ...
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug c++/105994] Incorrect optimization with -ftree-sra and casting (strictly compatible) refs
2022-06-16 0:10 [Bug c++/105994] New: Incorrect optimization with -ftree-sra and casting (strictly compatible) refs gcc_bugzilla.calin79 at neverbox dot com
2022-06-16 0:21 ` [Bug c++/105994] " pinskia at gcc dot gnu.org
2022-06-16 0:22 ` pinskia at gcc dot gnu.org
@ 2022-06-16 0:23 ` pinskia at gcc dot gnu.org
2022-06-16 0:25 ` pinskia at gcc dot gnu.org
2022-06-16 0:28 ` pinskia at gcc dot gnu.org
4 siblings, 0 replies; 6+ messages in thread
From: pinskia at gcc dot gnu.org @ 2022-06-16 0:23 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105994
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |RESOLVED
Resolution|--- |INVALID
--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Actually it is invalid because the temp goes out of scope rather than aliasing
issues.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug c++/105994] Incorrect optimization with -ftree-sra and casting (strictly compatible) refs
2022-06-16 0:10 [Bug c++/105994] New: Incorrect optimization with -ftree-sra and casting (strictly compatible) refs gcc_bugzilla.calin79 at neverbox dot com
` (2 preceding siblings ...)
2022-06-16 0:23 ` pinskia at gcc dot gnu.org
@ 2022-06-16 0:25 ` pinskia at gcc dot gnu.org
2022-06-16 0:28 ` pinskia at gcc dot gnu.org
4 siblings, 0 replies; 6+ messages in thread
From: pinskia at gcc dot gnu.org @ 2022-06-16 0:25 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105994
--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
const bar& test = h(foo{t})
The temporary for foo here since it is not bound directly to the variable bar,
goes out of scope after the statement is finished.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug c++/105994] Incorrect optimization with -ftree-sra and casting (strictly compatible) refs
2022-06-16 0:10 [Bug c++/105994] New: Incorrect optimization with -ftree-sra and casting (strictly compatible) refs gcc_bugzilla.calin79 at neverbox dot com
` (3 preceding siblings ...)
2022-06-16 0:25 ` pinskia at gcc dot gnu.org
@ 2022-06-16 0:28 ` pinskia at gcc dot gnu.org
4 siblings, 0 replies; 6+ messages in thread
From: pinskia at gcc dot gnu.org @ 2022-06-16 0:28 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105994
--- Comment #6 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Note with -fsanitize=address at runtime, GCC does detect this:
Program returned: 1
=================================================================
==1==ERROR: AddressSanitizer: stack-use-after-scope on address 0x7fe456600020
at pc 0x0000004011e4 bp 0x7ffee65e3850 sp 0x7ffee65e3848
READ of size 4 at 0x7fe456600020 thread T0
#0 0x4011e3 in g(foo) /app/example.cpp:20
#1 0x401209 in main /app/example.cpp:28
#2 0x7fe458c880b2 in __libc_start_main
(/lib/x86_64-linux-gnu/libc.so.6+0x240b2) (BuildId:
9fdb74e7b217d06c93172a8243f8547f947ee6d1)
#3 0x4010ad in _start (/app/output.s+0x4010ad) (BuildId:
5bbb0ffd1be8c2b44c3b09d7f1bcf41dcd821425)
Address 0x7fe456600020 is located in stack of thread T0 at offset 32 in frame
#0 0x401175 in g(foo) /app/example.cpp:17
This frame has 1 object(s):
[32, 36) '<unknown>' <== Memory access at offset 32 is inside this variable
HINT: this may be a false positive if your program uses some custom stack
unwind mechanism, swapcontext or vfork
(longjmp and C++ exceptions *are* supported)
SUMMARY: AddressSanitizer: stack-use-after-scope /app/example.cpp:20 in g(foo)
Shadow bytes around the buggy address:
0x0ffd0acb7fb0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x0ffd0acb7fc0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x0ffd0acb7fd0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x0ffd0acb7fe0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x0ffd0acb7ff0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
=>0x0ffd0acb8000: f1 f1 f1 f1[f8]f3 f3 f3 00 00 00 00 00 00 00 00
0x0ffd0acb8010: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x0ffd0acb8020: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x0ffd0acb8030: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x0ffd0acb8040: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
0x0ffd0acb8050: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
Shadow byte legend (one shadow byte represents 8 application bytes):
Addressable: 00
Partially addressable: 01 02 03 04 05 06 07
Heap left redzone: fa
Freed heap region: fd
Stack left redzone: f1
Stack mid redzone: f2
Stack right redzone: f3
Stack after return: f5
Stack use after scope: f8
Global redzone: f9
Global init order: f6
Poisoned by user: f7
Container overflow: fc
Array cookie: ac
Intra object redzone: bb
ASan internal: fe
Left alloca redzone: ca
Right alloca redzone: cb
==1==ABORTING
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2022-06-16 0:28 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-06-16 0:10 [Bug c++/105994] New: Incorrect optimization with -ftree-sra and casting (strictly compatible) refs gcc_bugzilla.calin79 at neverbox dot com
2022-06-16 0:21 ` [Bug c++/105994] " pinskia at gcc dot gnu.org
2022-06-16 0:22 ` pinskia at gcc dot gnu.org
2022-06-16 0:23 ` pinskia at gcc dot gnu.org
2022-06-16 0:25 ` pinskia at gcc dot gnu.org
2022-06-16 0:28 ` pinskia at gcc dot gnu.org
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).