Forums before death by AOL, social media and spammers... "We can't have nice things"
|    linux.debian.bugs.dist    |    Ohh some weird Debian bug report thing    |    28,835 messages    |
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
|    Message 27,279 of 28,835    |
|    Aurelien Jarno to All    |
|    Bug#1127698: djbdns: autopkgtest failure    |
|    11 Feb 26 23:10:02    |
   
   XPost: linux.debian.ports.arm   
   From: aurel32@debian.org   
      
   Source: djbdns   
   Version: 1:1.05-22.1   
   Severity: important   
   Tags: upstream patch   
   X-Debbugs-Cc: debian-arm@lists.debian.org   
   User: debian-arm@lists.debian.org   
   Usertags: arm64   
      
   Dear maintainer,   
      
   djbdns autopkgtest fails on arm64 when run against glibc 2.43, currently   
   in experimental. From the autopkgtest log:   
      
   | 37s ==== test_tinydns_run_udp   
   | 37s - spawned process 3012   
   | 37s - waiting for the 'starting tinydns' line   
   | 37s - got line 'starting tinydns\n'   
   | 37s - tinydns is listening at 127.6.16.1:53   
   | 37s - query packet: b'\x00\x11\x00\x00\x00\x01\x00\x00\x00\x00   
   x00\x00\x07example\x03com\x00\x00\x01\x00\x01'   
   | 37s - we spawned a process, checking if it has exited   
   | 37s Terminating process 3012   
   | 38s Query 8: expected a '+' in the tinydns log line '7f000001:e214:0018 -   
   0001 example.com\n'   
   | 38s autopkgtest [16:09:28]: test tinytest: -----------------------]   
      
   https://ci.debian.net/data/autopkgtest/unstable/arm64/d/djbdns/68572095/log.gz   
      
   After investigation, the failure is due to 2MB THP being enabled by   
   default in glibc 2.43 on arm64. This increases the size of the data   
   segment, causing it to exceed the soft limit defined by tinydns-conf.   
      
   The issue can also be reproduced with glibc 2.42, even on amd64, by   
   setting the GLIBC_TUNABLES environment variable to   
   glibc.malloc.hugetlb=2.   
      
   A possible way to fix the issue is to increase the data segment limit.   
   It seems that raising it from 7000000 (or 3000000 before #996807) to   
   13000000 is enough. Tested with the corresponding patch:   
      
   --- djbdns-1.05/debian/patches/0011-datalimit.patch   
   +++ djbdns-1.05/debian/patches/0011-datalimit.patch   
   @@ -10,7 +10,7 @@   
    start("run");   
    outs("#!/bin/sh\nexec 2>&1\nexec envdir ./env sh -c '\n exec envuidgid   
   "); outs(user);   
    - outs(" softlimit -d300000 tcpserver -vDRHl0 -x tcp.cdb -- \"$IP\" 53 ");   
   -+ outs(" softlimit -d700000 tcpserver -vDRHl0 -x tcp.cdb -- \"$IP\" 53 ");   
   ++ outs(" softlimit -d1300000 tcpserver -vDRHl0 -x tcp.cdb -- \"$IP\" 53 ");   
    outs(auto_home); outs("/sbin/axfrdns\n'\n");   
    finish();   
    perm(0755);   
   @@ -21,7 +21,7 @@   
    seed_addtime(); start("env/CACHESIZE"); outs("1000000\n"); finish();   
    seed_addtime(); perm(0644);   
    - seed_addtime(); start("env/DATALIMIT"); outs("3000000\n"); finish();   
   -+ seed_addtime(); start("env/DATALIMIT"); outs("7000000\n"); finish();   
   ++ seed_addtime(); start("env/DATALIMIT"); outs("13000000\n"); finish();   
    seed_addtime(); perm(0644);   
    seed_addtime(); start("run");   
    outs("#!/bin/sh\nexec 2>&1\nexec
|
[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]
(c) 1994, bbs@darkrealms.ca