cafkafk/nixpkgs
cafkafk/nixpkgs
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 2

pkgsi686Linux.samba: don't configure waf in parallel on 32-bit systems

Browse source 
Without the change `waf configure` hung on `i686Linux.samba` on systems
with large amount of host CPUs (for me it's 16).

This happens because one of the worker processes gets `-ENOMEM` and does
not recover from it:

    2084476 mmap2(NULL, 8392704, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_STACK, -1, 0) = -1 ENOMEM (Cannot allocate memory)
    2084476 munmap(0xf2b5c000, 16384)       = 0
    2084476 rt_sigprocmask(SIG_BLOCK, ~[RT_1], NULL, 8) = 0
    2084476 madvise(0x1ff000, 8372224, MADV_DONTNEED) = 0
    2084476 exit(0)                         = ?
    2084476 +++ exited with 0 +++

THe change extends 32-bit arm workaround to limit to one thread to all
32-bit systems.
This commit is contained in:
Sergei Trofimovich 2024-02-16 23:14:39 +00:00
parent 571716ebca
commit 9e38b7b0fb
1 changed files with 8 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
pkgs/servers/samba/4.x.nix
View file

@ -165,8 +165,14 @@ stdenv.mkDerivation rec {
++ optional (!enablePam) "--without-pam"
++ optionals (stdenv.hostPlatform != stdenv.buildPlatform) [
"--bundled-libraries=!asn1_compile,!compile_et"
] ++ optionals stdenv.isAarch32 [
# https://bugs.gentoo.org/683148
] ++ optionals stdenv.buildPlatform.is32bit [
# By default `waf configure` spawns as many as available CPUs. On
# 32-bit systems with many CPUs (like `i686` chroot on `x86_64`
# kernel) it can easily exhaust 32-bit address space and hang up:
# https://github.com/NixOS/nixpkgs/issues/287339#issuecomment-1949462057
# https://bugs.gentoo.org/683148
# Limit the job count down to the minimal on system with limited address
# space.
"--jobs 1"
];