public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
* [Bug target/56096] New: Bad code generated for conditional shift
@ 2013-01-24 6:47 tilman@code-monkey.de
2013-01-24 8:55 ` [Bug target/56096] " mikpe at it dot uu.se
` (4 more replies)
0 siblings, 5 replies; 6+ messages in thread
From: tilman@code-monkey.de @ 2013-01-24 6:47 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56096
Bug #: 56096
Summary: Bad code generated for conditional shift
Classification: Unclassified
Product: gcc
Version: 4.7.2
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: target
AssignedTo: unassigned@gcc.gnu.org
ReportedBy: tilman@code-monkey.de
Compiling this snippet
unsigned f1 (unsigned x, unsigned m)
{
x >>= ((m & 0x008080) ? 8 : 0);
return x;
}
with gcc 4.7.2 gives this code for ARMv5:
$ armv5tel-softfloat-linux-gnueabi-gcc -O2 -S -o- f.c
[...]
ldr r3, .L4
and r3, r1, r3
cmp r3, #0
movne r3, #8 @ XXX
moveq r3, #0 @ XXX
mov r0, r0, lsr r3 @ XXX
bx lr
[...]
Those three mov instructions are clearly sub-optimal.
Replacing the ternary operator with an if-statement gives the expected code
sequence:
unsigned f1 (unsigned x, unsigned m)
{
if (m & 0x008080)
x >>= 8;
return x;
}
->
ldr r3, .L6
and r3, r1, r3
cmp r3, #0
movne r0, r0, lsr #8
bx lr
ie we saved two mov instructions.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug target/56096] Bad code generated for conditional shift
2013-01-24 6:47 [Bug target/56096] New: Bad code generated for conditional shift tilman@code-monkey.de
@ 2013-01-24 8:55 ` mikpe at it dot uu.se
2013-03-05 9:06 ` [Bug tree-optimization/56096] Sub-optimal " rearnsha at gcc dot gnu.org
` (3 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: mikpe at it dot uu.se @ 2013-01-24 8:55 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56096
--- Comment #1 from Mikael Pettersson <mikpe at it dot uu.se> 2013-01-24 08:54:44 UTC ---
"Bad" is ambiguous, it could mean "sub-optimal" or it could mean "incorrect" or
"wrong". In this case it means "sub-optimal", please change the PR summary to
reflect that.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug tree-optimization/56096] Sub-optimal code generated for conditional shift
2013-01-24 6:47 [Bug target/56096] New: Bad code generated for conditional shift tilman@code-monkey.de
2013-01-24 8:55 ` [Bug target/56096] " mikpe at it dot uu.se
@ 2013-03-05 9:06 ` rearnsha at gcc dot gnu.org
2021-07-25 2:08 ` pinskia at gcc dot gnu.org
` (2 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: rearnsha at gcc dot gnu.org @ 2013-03-05 9:06 UTC (permalink / raw)
To: gcc-bugs
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56096
Richard Earnshaw <rearnsha at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Last reconfirmed| |2013-03-05
Component|target |tree-optimization
Ever Confirmed|0 |1
--- Comment #2 from Richard Earnshaw <rearnsha at gcc dot gnu.org> 2013-03-05 09:06:22 UTC ---
Confirmed. This is a generic problem with the way the tree optimizers and
expand work.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug tree-optimization/56096] Sub-optimal code generated for conditional shift
2013-01-24 6:47 [Bug target/56096] New: Bad code generated for conditional shift tilman@code-monkey.de
2013-01-24 8:55 ` [Bug target/56096] " mikpe at it dot uu.se
2013-03-05 9:06 ` [Bug tree-optimization/56096] Sub-optimal " rearnsha at gcc dot gnu.org
@ 2021-07-25 2:08 ` pinskia at gcc dot gnu.org
2023-08-05 3:49 ` pinskia at gcc dot gnu.org
2023-08-05 3:54 ` pinskia at gcc dot gnu.org
4 siblings, 0 replies; 6+ messages in thread
From: pinskia at gcc dot gnu.org @ 2021-07-25 2:08 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56096
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Target|armv5 |
Last reconfirmed|2013-03-05 00:00:00 |2021-7-24
Severity|normal |enhancement
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug tree-optimization/56096] Sub-optimal code generated for conditional shift
2013-01-24 6:47 [Bug target/56096] New: Bad code generated for conditional shift tilman@code-monkey.de
` (2 preceding siblings ...)
2021-07-25 2:08 ` pinskia at gcc dot gnu.org
@ 2023-08-05 3:49 ` pinskia at gcc dot gnu.org
2023-08-05 3:54 ` pinskia at gcc dot gnu.org
4 siblings, 0 replies; 6+ messages in thread
From: pinskia at gcc dot gnu.org @ 2023-08-05 3:49 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56096
--- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
GCC 7-10 produced:
movw r3, #32896
tst r1, r3
it ne
lsrne r0, r0, #8
bx lr
But GCC 11 produces:
movw r3, #32896
tst r1, r3
ite ne
movne r3, #1
moveq r3, #0
lsls r3, r3, #3
lsrs r0, r0, r3
Which is definitely worse.
^ permalink raw reply [flat|nested] 6+ messages in thread
* [Bug tree-optimization/56096] Sub-optimal code generated for conditional shift
2013-01-24 6:47 [Bug target/56096] New: Bad code generated for conditional shift tilman@code-monkey.de
` (3 preceding siblings ...)
2023-08-05 3:49 ` pinskia at gcc dot gnu.org
@ 2023-08-05 3:54 ` pinskia at gcc dot gnu.org
4 siblings, 0 replies; 6+ messages in thread
From: pinskia at gcc dot gnu.org @ 2023-08-05 3:54 UTC (permalink / raw)
To: gcc-bugs
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56096
--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
I think one way to improve the constant forming is during isel if we have
condition exec (e.g. arm or ia64) we should transform:
_7 = _1 != 0;
_8 = (int) _7;
_9 = _8 << 3;
to:
_9 = _1 ? 8 : 0;
That will at least fix this part:
movne r3, #1
moveq r3, #0
lsls r3, r3, #3
To fix the other part, I don't know ...
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2023-08-05 3:54 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2013-01-24 6:47 [Bug target/56096] New: Bad code generated for conditional shift tilman@code-monkey.de
2013-01-24 8:55 ` [Bug target/56096] " mikpe at it dot uu.se
2013-03-05 9:06 ` [Bug tree-optimization/56096] Sub-optimal " rearnsha at gcc dot gnu.org
2021-07-25 2:08 ` pinskia at gcc dot gnu.org
2023-08-05 3:49 ` pinskia at gcc dot gnu.org
2023-08-05 3:54 ` 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).