[Bug ipa/100812] [10 Regression] lto1-wpa memory hog building openjdk trunk
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100812 Richard Biener changed: What|Removed |Added Status|WAITING |RESOLVED Resolution|--- |FIXED Target Milestone|10.5|11.0 --- Comment #6 from Richard Biener --- Use GCC 11+
[Bug ipa/100812] [10 Regression] lto1-wpa memory hog building openjdk trunk
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100812 Jakub Jelinek changed: What|Removed |Added Target Milestone|10.4|10.5 --- Comment #5 from Jakub Jelinek --- GCC 10.4 is being released, retargeting bugs to GCC 10.5.
[Bug ipa/100812] [10 Regression] lto1-wpa memory hog building openjdk trunk
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100812
Martin Liška changed:
What|Removed |Added
Ever confirmed|0 |1
Status|UNCONFIRMED |WAITING
Last reconfirmed||2021-05-31
--- Comment #4 from Martin Liška ---
I recommend doing --save-temps --verbose and then running WPA without -quiet
flag. You'll see something like:
$ /home/marxin/bin/gcc/lib/gcc/x86_64-pc-linux-gnu/12.0.0/lto1 -dumpbase
./postgres.wpa -mtune=generic -march=x86-64 -O2 -O2 -O2 -Werror=vla
-Wimplicit-fallthrough=3 -Wformat-truncation=0 -Wno-stringop-truncation
-version -fno-openmp -fno-openacc -fno-pie -fcf-protection=none
-fno-strict-aliasing -fwrapv -fexcess-precision=standard
-fltrans-output-list=./postgres.ltrans.out -fwpa=16 -fresolution=postgres.res
-flinker-output=exec @./postgres.wpa.args.0
...
{heap 66M} {heap 66M} {heap 66M} {heap 66M}
{heap 66M} {GC released 3444k madv_dontneed 72k} {GC 143M
-> 128M} {heap 74M} {heap 74M} {heap 74M}
{GC released 1024k madv_dontneed 14M} {GC trimmed to 123M,
139M mapped} {heap 74M} {heap 74M} {heap 74M}
{heap 74M}
...
so one can identify the memory peak.
[Bug ipa/100812] [10 Regression] lto1-wpa memory hog building openjdk trunk
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100812 --- Comment #3 from Matthias Klose --- adding -fno-ipa-icf or -fno-ipa-sra, or both flags makes a difference
[Bug ipa/100812] [10 Regression] lto1-wpa memory hog building openjdk trunk
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100812 Richard Biener changed: What|Removed |Added Keywords||lto Component|lto |ipa Target Milestone|--- |10.4 Known to work||11.1.0 --- Comment #2 from Richard Biener --- Not sure if really a GCC regression - unless say GCC 9 is fine as well. That said, awaits reproduction / testcase. I suppose trying -fno-ipa-* to see whether triggered by a specific IPA pass would be interesting (my bet is ICF or SRA both of which have seen larger changes in GCC 10/11).
