Hi, 在 2026/6/23 19:30, Colin Watson 写道: > Source: nginx > Version: 1.30.1-4 > Severity: normal > X-Debbugs-Cc: [email protected], > [email protected] > User: [email protected] > Usertags: loong64 > > I noticed that debusine's autopkgtests are failing on loong64 (e.g. > https://ci.debian.net/packages/d/debusine/testing/loong64/72375884/). > On investigation, I found this in its test artifacts: > > Jun 22 10:47:53 ci-173-ef19176d nginx[8035]: 2026/06/22 10:47:53 [emerg] > 8035#8035: could not build server_names_hash, you should increase > server_names_hash_bucket_size: 32 > Jun 22 10:47:53 ci-173-ef19176d nginx[8035]: nginx: configuration file > /etc/nginx/nginx.conf test failed > > I suppose we could work around this in debusine, as suggested in > https://nginx.org/en/docs/http/server_names.html#optimization. However, > I think this message points to nginx being configured suboptimally on > this architecture. The default value of server_names_hash_bucket_size > is set to nginx's idea of the CPU cache line size. Looking at > https://salsa.debian.org/nginx-team/nginx/-/blob/debian/latest/auto/os/conf, > it seems that this is set based on matching the architecture name. No > specific value is set for the case where $NGX_MACHINE is loongarch64, so > it falls back to 32. Your investigation is appreciate!
> > I'm not an architecture expert, but looking at e.g. > https://github.com/golang/go/blob/master/src/internal/cpu/cpu_loong64.go > it seems as though 64 might be a better choice on this architecture. > Could anyone on debian-loongarch@ confirm or contradict this? Yes, 64 is a better choice on loong64 architecture. The following is run on my test loong64 machine: root@klx:~# arch loongarch64 root@klx:~# uname -a Linux klx 7.0.12+deb14-loong64 #1 SMP PREEMPT Debian 7.0.12-1 (2026-06-09) loongarch64 GNU/Linux root@klx:~# lscpu Architecture: loongarch64 CPU op-mode(s): 32-bit, 64-bit Address sizes: 47 bits physical, 48 bits virtual Byte Order: Little Endian CPU(s): 8 On-line CPU(s) list: 0-7 Model name: Loongson-3A5000 CPU family: Loongson-64bit Model: 0x10 Thread(s) per core: 1 Core(s) per socket: 8 Socket(s): 1 BogoMIPS: 4000.00 Flags: cpucfg lam ual fpu lsx lasx crc32 ptw lspw lbt_x86 lbt_arm lbt_mips Caches (sum of all): L1d: 512 KiB (8 instances) L1i: 512 KiB (8 instances) L2: 2 MiB (8 instances) L3: 16 MiB (1 instance) NUMA: NUMA node(s): 1 NUMA node0 CPU(s): 0-7 Vulnerabilities: Gather data sampling: Not affected Ghostwrite: Not affected Indirect target selection: Not affected Itlb multihit: Not affected L1tf: Not affected Mds: Not affected Meltdown: Not affected Mmio stale data: Not affected Old microcode: Not affected Reg file data sampling: Not affected Retbleed: Not affected Spec rstack overflow: Not affected Spec store bypass: Not affected Spectre v1: Mitigation; __user pointer sanitization Spectre v2: Not affected Srbds: Not affected Tsa: Not affected Tsx async abort: Not affected Vmscape: Not affected root@klx:~# cat /sys/devices/system/cpu/cpu0/cache/index0/coherency_line_size 64 I also confirm this through the Chinese Wechat group with staff in loongson company. Regards, -- 肖盛文 xiao sheng wen -- Debian Developer(atzlinux) Debian QA page: https://qa.debian.org/developer.php?login=atzlinux%40debian.org Debian salsa: https://salsa.debian.org/atzlinux-guest GnuPG Public Key: 0x00186602339240CB
OpenPGP_0x00186602339240CB.asc
Description: OpenPGP public key
OpenPGP_signature.asc
Description: OpenPGP digital signature
