Content-Type: text/plain; charset=utf-8
Precedence: list
Mime-Version: 1.0 (Mac OS X Mail 9.2 \(3112\))
Subject: Re: [alpine-aports] [PATCH] testing/mongodb: fix for #5117
From: Marc Vertes <marc.vertes@ugrid.net>
In-Reply-To: <20160224084437.2b11cbe1@vostro.util.wtbts.net>
Date: Wed, 24 Feb 2016 10:32:59 +0100
Cc: alpine-aports@lists.alpinelinux.org
Content-Transfer-Encoding: quoted-printable
Message-Id: <7AA15BC7-7B5F-4973-ADEC-6ADCADECD11F@ugrid.net>
References: <1456247469-958-1-git-send-email-marc.vertes@ugrid.net> <20160223191612.6f25f25b@vostro> <6B28E7B4-454F-4391-AA4E-498D73DD811A@ugrid.net> <C0783129-1A80-4D3B-8D2F-6D76A733095E@ugrid.net> <20160223194440.3fc44b9a@vostro> <2309A9D8-2394-4092-8BF5-BA4E3C13870F@ugrid.net> <20160224084437.2b11cbe1@vostro.util.wtbts.net>
To: Timo Teras <timo.teras@iki.fi>


> Le 24 f=C3=A9vr. 2016 =C3=A0 07:44, Timo Teras <timo.teras@iki.fi> a =
=C3=A9crit :
>=20
> On Tue, 23 Feb 2016 23:46:39 +0100
> Marc Vertes <marc.vertes@ugrid.net> wrote:
>=20
>>> Le 23 f=C3=A9vr. 2016 =C3=A0 18:44, Timo Teras <timo.teras@iki.fi> a =
=C3=A9crit :
>>>=20
>>> So the current patch is wrong. pthread_getattr_np returns valid and
>>> correct data for non-main threads. Only main thread should get the
>>> "treatment".
>>>=20
>> Ok, I agree, there is no need to patch the reporting of stack base /
>> size which are correct (obtained by pthread_getattr_np in a non-main
>> thread).
>>=20
>> So the real problem comes from the small default stack size (80k)
>> imposed by musl.
>>=20
>> I saw that reducing the quota from 64k to 32k in
>> StackLocator.available() which checks remaining stack space, was
>> enough to avoid the initial crash.
>=20
> Should we start by applying that one? At least to make mongdb compile
> again as my change broke the build (bad checksums).

I'm preparing a new patch, removing musl-process-stack-size.patch and
adjusting the check of minimum available stack space from 64k to 32k.
 =20
>=20
>> So the normal thing to do would be increase a bit the stack space for
>> created threads.
>>=20
>> I can't use pthread_attr_setstacksize before pthread_create, because
>> of the use of C++ threads (what a pile of crap!). I tried to force a
>> non-zero stacksize in PR_CreateThread(), but it seems to do nothing.
>=20
> Urgh. Yeah, that looks abhorrent. I'll ping #musl what could be done.
>=20
>> The 2 last remaining options: test if g++ -fsplit-stack would work,
>> or reduce quota as mentioned before.
>>=20
>> What do you think ?
>=20
> split-stack has it's own problems too. IIRC, it breaks other stuff. =
And
> requires gold linker to even work (we have it, but do not use it be
> default).
>=20
Ok, I won't proceed further on that.

> glibc ships pthread_setattr_default_np() that can be used to change =
the
> default stack size. I wonder if some similar mechanism could be added
> to musl.
>=20
It would be certainly useful to ease porting multi-threaded C++ stuff =
needing
big stack space for their threads. It's not because alpine and musl
are concise and elegant that they should not support big bloated craps.

> But as said, maybe start with changing the marginal from 64->32kB?
> In musl the stack size does not contain the guard page (which is in
> agreement with posix's definition of thread attribute stack =
base/size).
> The comment did say the check is not done often enough though, so that
> might not be desirable long term solution. And that should probably be
> explained in the patch.
>=20
Yes. Will do that.

Thanks,
Marc


---
Unsubscribe:  alpine-aports+unsubscribe@lists.alpinelinux.org
Help:         alpine-aports+help@lists.alpinelinux.org
---