[Bug bootstrap/69709] [6 Regression] profiled bootstrap error on s390x-linux-gnu with r233194
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69709 Jakub Jelinek changed: What|Removed |Added Status|WAITING |RESOLVED Resolution|--- |FIXED --- Comment #15 from Jakub Jelinek --- Hopefully fixed then.
[Bug bootstrap/69709] [6 Regression] profiled bootstrap error on s390x-linux-gnu with r233194
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69709 --- Comment #14 from Andreas Krebbel --- Author: krebbel Date: Fri Feb 26 18:03:51 2016 New Revision: 233752 URL: https://gcc.gnu.org/viewcvs?rev=233752&root=gcc&view=rev Log: S/390: PR69709 Fix risbg splitter This fixes a wrong code generation problem with the splitters introduced with that patch: https://gcc.gnu.org/ml/gcc-patches/2015-07/msg01840.html The target operand is used as temporary. This fails if it matches the source of the left shift which is read after writing the temporary. Thanks to Dominik for debugging it and thanks to Richard for the fix! Bootstrapped and regtested on s390x with-arch=z13. Bye, -Andreas- gcc/ChangeLog: 2016-02-26 Richard Henderson PR target/69709 * config/s390/s390.md (risbg and risbgn splitters): Allocate new pseudo in case the target rtx matches the source of the left shift. gcc/testsuite/ChangeLog: 2016-02-26 Andreas Krebbel PR target/69709 * gcc.target/s390/pr69709.c: New test. Added: trunk/gcc/testsuite/gcc.target/s390/pr69709.c Modified: trunk/gcc/ChangeLog trunk/gcc/config/s390/s390.md trunk/gcc/testsuite/ChangeLog
[Bug bootstrap/69709] [6 Regression] profiled bootstrap error on s390x-linux-gnu with r233194
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69709 --- Comment #13 from Andreas Krebbel --- Yes. This was problem. Your patch fixes this. Thanks!
[Bug bootstrap/69709] [6 Regression] profiled bootstrap error on s390x-linux-gnu with r233194
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69709 --- Comment #12 from Richard Henderson --- Created attachment 37799 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=37799&action=edit example patch Or, for the purposes of combine, better like so. This avoids creating a third (move) insn, which will foil the 3->2 insn combination.
[Bug bootstrap/69709] [6 Regression] profiled bootstrap error on s390x-linux-gnu with r233194
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69709
Richard Henderson changed:
What|Removed |Added
CC||rth at gcc dot gnu.org
--- Comment #11 from Richard Henderson ---
Ho hum. I've been looking at this this afternoon too.
Just to be sure we're on the same page, it's the insv
splitters not handling overlapping inputs.
diff --git a/gcc/config/s390/s390.md b/gcc/config/s390/s390.md
index 2c90eae..a3f3f0a 100644
--- a/gcc/config/s390/s390.md
+++ b/gcc/config/s390/s390.md
@@ -3883,6 +3883,12 @@
(ashift:GPR (match_dup 3) (match_dup 4]
{
operands[5] = GEN_INT ((1UL << UINTVAL (operands[4])) - 1);
+ if (rtx_equal_p (operands[0], operands[3]))
+{
+ if (!can_create_pseudo_p ())
+ FAIL;
+ operands[3] = copy_to_reg (operands[3]);
+}
})
(define_split
@@ -3903,6 +3909,12 @@
(clobber (reg:CC CC_REGNUM))])]
{
operands[5] = GEN_INT ((1UL << UINTVAL (operands[4])) - 1);
+ if (rtx_equal_p (operands[0], operands[3]))
+{
+ if (!can_create_pseudo_p ())
+ FAIL;
+ operands[3] = copy_to_reg (operands[3]);
+}
})
(define_insn "*rsbg__noshift"
[Bug bootstrap/69709] [6 Regression] profiled bootstrap error on s390x-linux-gnu with r233194
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69709 --- Comment #10 from Dominik Vogt --- We've located the bug in the s390 backend. No further help is needed.
[Bug bootstrap/69709] [6 Regression] profiled bootstrap error on s390x-linux-gnu with r233194
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69709 --- Comment #9 from Dominik Vogt --- (-fpeel-loops is activated by -fprofile-use, so this is the connection to profilesbootstrap.)
[Bug bootstrap/69709] [6 Regression] profiled bootstrap error on s390x-linux-gnu with r233194
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69709 --- Comment #8 from Dominik Vogt --- Created attachment 37790 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=37790&action=edit Test case The option -fpeel-loops triggers the bug. The attached program has a different result with -fpeel-loops than without it. $ gcc -O2 -march=z10 -fpeel-loops pr69709.c && ./a.out 1 bits set in result $ gcc -O2 -march=z10 pr69709.c && ./a.out 2 bits set in result (2 is the correct result).
[Bug bootstrap/69709] [6 Regression] profiled bootstrap error on s390x-linux-gnu with r233194
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69709
--- Comment #7 from Dominik Vogt ---
The stage1 compiler does something wrong when compiling gcc/real.c (with
-fprofile-generate). The function div_significands() (inlined into
do_divide()) returns a wrong result due to bad register usage in this loop:
-- snip --
do
{
msb = u.sig[SIGSZ-1] & SIG_MSB;
lshift_significand_1 (&u, &u);
start:
if (msb || cmp_significands (&u, b) >= 0)
{
sub_significands (&u, &u, b, 0);
set_significand_bit (r, bit);
}
}
while (--bit >= 0);
-- snip --
At loop entry ("start" label), r1 holds the highest 64 bits of the significand.
The first pass through the loop seems to be correct; sub_significands() and
set_significand_bit() do the correct operations. After that, r1 is decremented
by one as if it contained the variable "bit". Later on r1 gets (eventually)
overwritten with zero. After that, the loop always thinks that the remaining
significand is too smaller than b because its always zero.
In the end, the "result" of the division is one in the highest significand bit
and all other bits zero, eventually causing the observed assertion failure.
With a broken compiler (from stageprofile), the test program for triggering the
ICE is simply
-- snip --
int x = __DBL_MAX__;
-- snip --
All of this only happens on a Fedora 20 chroot for me. I've tried to add
"-save-temps -dA -dP -fdump-rtl-all" to the OPT_FLAGS in the rpm spec file, but
then the package doesn't build at all. Any hints how to get debug information
from the rembuild run is welcome.
[Bug bootstrap/69709] [6 Regression] profiled bootstrap error on s390x-linux-gnu with r233194
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69709 --- Comment #6 from Jakub Jelinek --- FYI, profiledbootstrap works for me including Ada, see http://s390.koji.fedoraproject.org/packages/gcc/6.0.0/0.12.fc24/data/logs/s390x/build.log
[Bug bootstrap/69709] [6 Regression] profiled bootstrap error on s390x-linux-gnu with r233194
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69709 --- Comment #5 from Dominik Vogt --- @Matthias: So far it only happens for me when building a gcc rpm from source on a (very slow VM), but not when compiling the same sources. Is there anything special about your build machine or environment on it?
[Bug bootstrap/69709] [6 Regression] profiled bootstrap error on s390x-linux-gnu with r233194
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69709 --- Comment #4 from Matthias Klose --- still seen today; normal bootstrap works.
[Bug bootstrap/69709] [6 Regression] profiled bootstrap error on s390x-linux-gnu with r233194
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69709 Dominik Vogt changed: What|Removed |Added CC||vogt at linux dot vnet.ibm.com --- Comment #3 from Dominik Vogt --- Building the source Rpm you sent us (to build the Ada compiler) has the same problem with profiledbootstrap, building on Fedora 20 (maybe other distros too). I'll try to isolate the problem.
[Bug bootstrap/69709] [6 Regression] profiled bootstrap error on s390x-linux-gnu with r233194
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69709 Jakub Jelinek changed: What|Removed |Added Status|UNCONFIRMED |WAITING Last reconfirmed||2016-02-17 Ever confirmed|0 |1 --- Comment #2 from Jakub Jelinek --- Can you still reproduce this? I've succeeded with several profiledbootstraps on s390x-linux as well as s390-linux.
[Bug bootstrap/69709] [6 Regression] profiled bootstrap error on s390x-linux-gnu with r233194
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69709 Jakub Jelinek changed: What|Removed |Added CC||jakub at gcc dot gnu.org --- Comment #1 from Jakub Jelinek --- Strange, profiledbootstrap worked just fine for me during the weekend on s390x-linux-gnu, but (with r233209 applied by hand) I got profiledbootstrap failure on popwerpc64-linux-gnu (big endian) instead (ICE in reload).
[Bug bootstrap/69709] [6 Regression] profiled bootstrap error on s390x-linux-gnu with r233194
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69709 Richard Biener changed: What|Removed |Added Keywords||build Priority|P3 |P1 Target Milestone|--- |6.0
