public inbox for io-uring@vger.kernel.org
 help / color / mirror / Atom feed
From: Jens Axboe <axboe@kernel.dk>
To: "Milan P. Stanić" <mps@arvanta.net>
Cc: io-uring@vger.kernel.org
Subject: Re: Building liburing on musl libc gives error that errno.h not found
Date: Tue, 17 Jun 2025 05:51:36 -0600	[thread overview]
Message-ID: <13d5c94e-9f3c-4a0e-a468-562e78155ba5@kernel.dk> (raw)
In-Reply-To: <20250617081947.GA21947@m1pro.arvanta.net>

On 6/17/25 2:19 AM, Milan P. Stani? wrote:
> On Mon, 2025-06-16 at 10:35, Jens Axboe wrote:
>> On 6/16/25 10:32 AM, Milan P. Stani? wrote:
>>> On Mon, 2025-06-16 at 09:35, Jens Axboe wrote:
>>>> On 6/16/25 9:13 AM, Jens Axboe wrote:
>>>>> On 6/16/25 8:18 AM, Milan P. Stani? wrote:
>>>>>> On Mon, 2025-06-16 at 07:59, Jens Axboe wrote:
>>>>>>> On 6/16/25 7:06 AM, Milan P. Stani? wrote:
>>>>>>>> On Mon, 2025-06-16 at 06:34, Jens Axboe wrote:
>>>>>>>>> On 6/15/25 1:56 PM, Milan P. Stani? wrote:
>>>>>>>>>> On Sun, 2025-06-15 at 12:57, Jens Axboe wrote:
>>>>>>>>>>> On 6/15/25 11:16 AM, Milan P. Stani? wrote:
>>>>>>>>>>>> Hi,
>>>>>>>>>>>>
>>>>>>>>>>>> Trying to build liburing 2.10 on Alpine Linux with musl libc got error
>>>>>>>>>>>> that errno.h is not found when building examples/zcrx.c
>>>>>>>>>>>>
>>>>>>>>>>>> Temporary I disabled build zcrx.c, merge request with patch for Alpine
>>>>>>>>>>>> is here:
>>>>>>>>>>>> https://gitlab.alpinelinux.org/alpine/aports/-/merge_requests/84981
>>>>>>>>>>>> I commented in merge request that error.h is glibc specific.
>>>>>>>>>>>
>>>>>>>>>>> I killed it, it's not needed and should've been caught during review.
>>>>>>>>>>> We should probably have alpine/musl as part of the CI...
>>>>>>>>>>
>>>>>>>>>> Fine.
>>>>>>>>>>
>>>>>>>>>>>> Side note: running `make runtests` gives 'Tests failed (32)'. Not sure
>>>>>>>>>>>> should I post full log here.
>>>>>>>>>>>
>>>>>>>>>>> Either that or file an issue on GH. Sounds like something is very wrong
>>>>>>>>>>> on the setup if you get failing tests, test suite should generally
>>>>>>>>>>> pass on the current kernel, or any -stable kernel.
>>>>>>>>>>>
>>>>>>>>>> I'm attaching log here to this mail. Actually it is one bug but repeated
>>>>>>>>>> in different tests, segfaults
>>>>>>>>>
>>>>>>>>> Your kernel is ancient, and that will surely account from some of the
>>>>>>>>> failures you see. A 6.6 stable series from January 2024 is not current
>>>>>>>>> by any stretch, should definitely upgrade that. But I don't think this
>>>>>>>>> accounts for all the failures seen, it's more likely there's some musl
>>>>>>>>> related issue as well which is affecting some of the tests.
>>>>>>>>
>>>>>>>> This happens also on 6.14.8-1 asahi kernel on apple m1pro machine.
>>>>>>>> I forgot to mention this in previous mail, sorry.
>>>>>>>
>>>>>>> Also on musl, correct?
>>>>>>
>>>>>> Yes, correct.
>>>>>>
>>>>>>> Guessing it must be some musl oddity. I'll try and setup a vm with
>>>>>>> alpine and see how that goes.
>>>>>>
>>>>>> It could be. I can ask on #musl IRC channel on libera.chat
>>>>>
>>>>> Probably easier if I just take a look at it, as long as I can get
>>>>> an alpine vm image going.
>>>>
>>>> Pure guesswork, but you are most likely running into default ulimit
>>>> limits being tiny. Probably something ala:
>>>>
>>>> rc_ulimit="-n 524288 -l 262144"
>>>>
>>>> in /etc/rc.conf would help.
>>>
>>> Tried, but didn't help.
>>>
>>> I will left it for now and return to test it when new liburing is
>>> released. It must pass our builders and CI, so I disabled test earlier.
>>>
>>> Thank you for help.
>>
>> That's fine, I don't recommend distros attempt to verify it by using
>> the test suite anyway, that's not really its intended purpose. Though it
>> can be useful in terms of verifying all relevant fixes are backported,
>> particular if the distro is one of those oddballs that don't base on or
>> pull in -stable.
>>
>> I'll be releasing 2.11 shortly, but it likely won't change anything on
>> your end, outside of having the examples/zcrx compilation fixed.
>>
>> FWIW, I'm on Alpine Linux 3.22 and it passes here.
> 
> liburing 2.11 builds fine on Alpine edge and previous bugs (segfaults)
> don't appears now with new version.
> 
> runtests shows that 4 test failed:
> ----------------
> Test run complete, kernel: 6.6.14-0-lts #1-Alpine SMP PREEMPT_DYNAMIC Fri, 26 Jan 2024 11:08:07 +0000
> Tests failed (4): <accept.t> <nop.t> <sqwait.t> <timeout.t>
> make[1]: *** [Makefile:331: runtests] Error 1
> make[1]: Leaving directory '/home/mps/aports/main/liburing/src/liburing-liburing-2.11/test'
> make: *** [Makefile:21: runtests] Error 2
> ----------------
> 
> but I think this is not blocker to update liburing to 2.11 on alpine.

It's fine - as mentioned, by far most of the test suite is for kernel
testing, not liburing itself. And if run on an old kernel. various test
case failures are expected, as the kernel is lacking fixes.

-- 
Jens Axboe

      reply	other threads:[~2025-06-17 11:51 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-06-15 17:16 Building liburing on musl libc gives error that errno.h not found Milan P. Stanić
2025-06-15 18:57 ` Jens Axboe
2025-06-15 19:56   ` Milan P. Stanić
2025-06-16 12:34     ` Jens Axboe
2025-06-16 13:06       ` Milan P. Stanić
2025-06-16 13:59         ` Jens Axboe
2025-06-16 14:18           ` Milan P. Stanić
2025-06-16 15:13             ` Jens Axboe
2025-06-16 15:26               ` Jens Axboe
2025-06-16 15:35               ` Jens Axboe
2025-06-16 16:32                 ` Milan P. Stanić
2025-06-16 16:35                   ` Jens Axboe
2025-06-16 16:44                     ` Milan P. Stanić
2025-06-17  8:19                     ` Milan P. Stanić
2025-06-17 11:51                       ` Jens Axboe [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=13d5c94e-9f3c-4a0e-a468-562e78155ba5@kernel.dk \
    --to=axboe@kernel.dk \
    --cc=io-uring@vger.kernel.org \
    --cc=mps@arvanta.net \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox