Re: The 2.40 branch has been created

Re: The 2.40 branch has been created

[Mark Wielaard]

[-- Attachment #1: Type: text/plain, Size: 856 bytes --]

Hi Christophe,

On Tue, Jan 03, 2023 at 06:29:30PM +0100, Christophe Lyon wrote:
> Actually we have:
> armhf-ubuntu20_04 and armhf-ubuntu22_04 to build GDB and
> ubuntu22_04-armhf to build GCC
> 
> The containers we use to build GDB have less cores than the ones we
> use for GCC. If we add binutils builders on the same workers as GDB,
> will they cooperate nicely (as in "do not compete for CPU", ie. be
> scheduled at different times)?
> 
> If yes, maybe it makes sense to add binutils builders to:
> armhf-ubuntu20_04
> armhf-ubuntu22_04
> arm64-ubuntu20_04
> arm64-ubuntu22_04
> 
> WDYT?

That makes sense. Yes, they will cooperate nicely because the workers
all have max_builds=1. So if there are multiple builds that could use
the same worker they will just wait in the pending queue.

I'll push the following to try the new builders.

Thanks,

Mark

[-- Attachment #2: 0001-Add-binutils-ubuntu20_04-and-ubuntu22_04-armhf-and-a.patch --]
[-- Type: text/plain, Size: 2566 bytes --]

From e125fa6b21d87c8bb39649299762cdfbafe0282e Mon Sep 17 00:00:00 2001
From: Mark Wielaard <mark@klomp.org>
Date: Wed, 4 Jan 2023 01:16:18 +0100
Subject: [PATCH] Add binutils ubuntu20_04 and ubuntu22_04 armhf and arm64
 builders

---
 builder/master.cfg | 34 +++++++++++++++++++++++++++++++++-
 1 file changed, 33 insertions(+), 1 deletion(-)

diff --git a/builder/master.cfg b/builder/master.cfg
index 8a66ae8..b16bb29 100644
--- a/builder/master.cfg
+++ b/builder/master.cfg
@@ -702,7 +702,11 @@ binutils_scheduler = schedulers.SingleBranchScheduler(
                       "binutils-debian-i386",
                       "binutils-fedora-ppc64le",
                       "binutils-opensusetw-x86_64",
-                      "binutils-opensuseleap-x86_64"])
+                      "binutils-opensuseleap-x86_64",
+                      "binutils-ubuntu20_04-armhf",
+                      "binutils-ubuntu22_04-armhf",
+                      "binutils-ubuntu20_04-arm64",
+                      "binutils-ubuntu22_04-arm64"])
 c['schedulers'].append(binutils_scheduler)
 
 binutils_try_scheduler = schedulers.AnyBranchScheduler(
@@ -2331,6 +2335,34 @@ binutils_opensuseleap_x86_64_builder = util.BuilderConfig(
         factory=binutils_factory)
 c['builders'].append(binutils_opensuseleap_x86_64_builder)
 
+binutils_ubuntu20_04_armhf_builder = util.BuilderConfig(
+	name="binutils-ubuntu20_04-armhf",
+        workernames=["armhf-ubuntu20_04"],
+        tags=["binutils", "armhf", "ubuntu"],
+        factory=binutils_factory)
+c['builders'].append(binutils_ubuntu20_04_armhf_builder)
+
+binutils_ubuntu22_04_armhf_builder = util.BuilderConfig(
+	name="binutils-ubuntu22_04-armhf",
+        workernames=["armhf-ubuntu22_04"],
+        tags=["binutils", "armhf", "ubuntu"],
+        factory=binutils_factory)
+c['builders'].append(binutils_ubuntu22_04_armhf_builder)
+
+binutils_ubuntu20_04_arm64_builder = util.BuilderConfig(
+	name="binutils-ubuntu20_04-arm64",
+        workernames=["arm64-ubuntu20_04"],
+        tags=["binutils", "arm64", "ubuntu"],
+        factory=binutils_factory)
+c['builders'].append(binutils_ubuntu20_04_arm64_builder)
+
+binutils_ubuntu22_04_arm64_builder = util.BuilderConfig(
+	name="binutils-ubuntu22_04-arm64",
+        workernames=["arm64-ubuntu22_04"],
+        tags=["binutils", "arm64", "ubuntu"],
+        factory=binutils_factory)
+c['builders'].append(binutils_ubuntu22_04_arm64_builder)
+
 binutils_rawhide_x86_64_builder = util.BuilderConfig(
 	name="binutils-rawhide-x86_64",
 	properties={'container-file':
-- 
2.31.1

Re: The 2.40 branch has been created

[Christophe Lyon]

On 1/4/23 01:35, Mark Wielaard wrote:
> Hi Christophe,
> 
> On Tue, Jan 03, 2023 at 06:29:30PM +0100, Christophe Lyon wrote:
>> Actually we have:
>> armhf-ubuntu20_04 and armhf-ubuntu22_04 to build GDB and
>> ubuntu22_04-armhf to build GCC
>>
>> The containers we use to build GDB have less cores than the ones we
>> use for GCC. If we add binutils builders on the same workers as GDB,
>> will they cooperate nicely (as in "do not compete for CPU", ie. be
>> scheduled at different times)?
>>
>> If yes, maybe it makes sense to add binutils builders to:
>> armhf-ubuntu20_04
>> armhf-ubuntu22_04
>> arm64-ubuntu20_04
>> arm64-ubuntu22_04
>>
>> WDYT?
> 
> That makes sense. Yes, they will cooperate nicely because the workers
> all have max_builds=1. So if there are multiple builds that could use
> the same worker they will just wait in the pending queue.
> 
> I'll push the following to try the new builders.

Thanks!

I see they are all failing currently ;-)
I have sent patches for the armhf failures yesterday, so they will 
hopefully become green soon.
I don't understand why the arm64 ones are failing, is it because there 
is one linker test with XPASS? (I didn't notice any "unexpected failure" 
in the .sum summaries)

Christophe

> 
> Thanks,
> 
> Mark

Re: The 2.40 branch has been created

[Mark Wielaard]

Hi Christophe,

On Wed, Jan 04, 2023 at 11:31:50AM +0100, Christophe Lyon wrote:
> >I'll push the following to try the new builders.
> 
> Thanks!
> 
> I see they are all failing currently ;-)
> I have sent patches for the armhf failures yesterday, so they will
> hopefully become green soon.
> I don't understand why the arm64 ones are failing, is it because
> there is one linker test with XPASS? (I didn't notice any
> "unexpected failure" in the .sum summaries)

Yes, the "XPASS: Run pr19719 fun undefined" is unexpected and so fails
make check.

Note that same XPASS on binutils-fedora-arm64, but that also has an
extra FAIL: Build pr26094-1

Cheers,

Mark