Re: [Buildroot] [PATCH 2/3] support/tests: allow top-level parallel builds

["Yann E. MORIN" <yann.morin.1998@free.fr>]

Thomas, All,

On 2022-12-27 21:53 +0100, Yann E. MORIN spake thusly:
> On 2022-12-27 21:45 +0100, Thomas Petazzoni via buildroot spake thusly:
> > On Sat, 24 Dec 2022 10:18:12 +0100
> > "Yann E. MORIN" <yann.morin.1998@free.fr> wrote:
> > > Running tests with top-level parallel builds can speed up running some
> > > tests, expecially those that have a lot of packages like the systemd
> > > init tests.
> > > 
> > > Trigger TLPB when the configuration enables per-package directories.
> > > 
> > > Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
> > > Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
> > 
> > I certainly support the idea of running some tests with TLPB. However,
> > this implementation makes a confusion between two different settings:
> > 
> > - The existing BRConfigTest.jlevel, which is set by run-tests -j, and
> >   used to define BR2_JLEVEL in the Buildroot configuration of each test
> >   case. This determines the number of parallel jobs used to build each
> >   independent package.
> > 
> > - Your new work, which uses "make -j" to do TLPB... but relies on the
> >   same above value, even though it's a completely different setting.
> > 
> > Is this expected?
> 
> Yes, this is the intended behaviour, which I was explicitly seeking.
> 
> So, if one runs with PPD and TLPB (outside the run-time infra), one
> would do something like:
> 
>     $ make -jN
> 
> This spawns a top-level make process that is parallel. In turn, in
> rules, when we call $(MAKE), this is the magic that tells make that
> it is recursive, but that it should use the jobserver from the calling
> process.
> 
> So, in this case, the BR2_JLEVEL is unused by whatever uses the make
> jobserver; only the number of jobs in the top-level jobserver is
> meaningful, i.e. whatever we pass as -jN.

Slight correction: BR2_JLEVEL does have an actual effect, but only if its
value is lower than the one we pass as -jN.

So, in practice, we want to have BR2_JLEVEL equal to the top-level -jN.

Regards,
Yann E. MORIN.

> The exception, of course, is whatever uses BR2_JLEVEL but does not talk
> to the jobserver, but this is mostly a few packages (scons, waf et al.).
> Even ninja packages do talk to the top-level jobserver, now that we use
> the ninja fork that knows to talk to it.
> 
> So, yes, using top-level -jN with the same value as BR2_JLEVEL is
> exactly what I intended to do.
> 
> Unless I totally missed something...
> 
> Regards,
> Yann E. MORIN.
> 
> -- 
> .-----------------.--------------------.------------------.--------------------.
> |  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
> | +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
> | +33 561 099 427 `------------.-------:  X  AGAINST      |  \e/  There is no  |
> | http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
> '------------------------------^-------^------------------^--------------------'
> _______________________________________________
> buildroot mailing list
> buildroot@buildroot.org
> https://lists.buildroot.org/mailman/listinfo/buildroot

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 561 099 427 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'
_______________________________________________
buildroot mailing list
buildroot@buildroot.org
https://lists.buildroot.org/mailman/listinfo/buildroot