libsframe builder (Was: [PATCH 0/2] libsframe: fix some memory leaks)
Checks
Commit Message
Hi Indu,
On Thu, 2022-12-22 at 14:54 -0800, Indu Bhagat via Binutils wrote:
> This patch set fixes some memory leaks in the libsframe and it's
> testsuite.
I see you are an enthusiastic user of the binutils-try buildbot. But
changes under libsframe/ don't trigger new builds and the libsframe
testsuite isn't actually ran.
The following tweak to the buildbot master.cfg should fix that.
Please let buildbot@sourceware.org know if you need more tweaks to
better test the new code.
Cheers,
Mark
Comments
Hi -
> command=['make',
> util.Interpolate('-j%(prop:ncpus)s'),
> - 'check-ld', 'check-gas', 'check-binutils'],
> + 'check-ld', 'check-gas', 'check-binutils',
> + 'check-libsframe'],
> name='make check',
> logfiles={ "ld.sum": "ld/ld.sum",
> "ld.log": "ld/ld.log",
> "gas.sum": "gas/testsuite/gas.sum",
> "gas.log": "gas/testsuite/gas.log",
> "binutils.sum": "binutils/binutils.sum",
> - "binutils.log": "binutils/binutils.log" },
> + "binutils.log": "binutils/binutils.log",
> + "libsframe.sum": "libsframe/libsframe.sum",
> + "libsframe.log": "libsframe/libsframe.log" },
> [...]
(For the record, all of those .log/.sum files, and more, are pulled
into the bunsen database automagically, so listing them all here for
storage and presentation in the buildbot system is not really
necessary.)
- FChE
On 12/23/22 06:22, Mark Wielaard wrote:
> Hi Indu,
>
> On Thu, 2022-12-22 at 14:54 -0800, Indu Bhagat via Binutils wrote:
>> This patch set fixes some memory leaks in the libsframe and it's
>> testsuite.
>
> I see you are an enthusiastic user of the binutils-try buildbot. But
> changes under libsframe/ don't trigger new builds and the libsframe
> testsuite isn't actually ran.
>
heh, binutils-try buildbot's enthusiastic user I am, yes. I did wonder
if there are enough resources for users to be doing this so frequently
(with almost every patch set before commit), I hope its not a problem ?
Right, I noticed that libsframe/ testsuite is not being run, but
binutils-try buildbot has still been useful to catch regressions to
gas/ld by my code ...
Asking about adding libsframe/ testsuite was on my list, thanks for
doing that.
Best Regards,
Indu
> The following tweak to the buildbot master.cfg should fix that.
>
> Please let buildbot@sourceware.org know if you need more tweaks to
> better test the new code.
>
> Cheers,
>
> Mark
Hi Frank,
On Fri, 2022-12-23 at 09:41 -0500, Frank Ch. Eigler wrote:
> (For the record, all of those .log/.sum files, and more, are pulled
> into the bunsen database automagically, so listing them all here for
> storage and presentation in the buildbot system is not really
> necessary.)
You are of course right. But the bunsen link is not easily found. We
really need to tweak the web-interface and emails sent to prominently
feature the bunsen results link.
They can be found in the stdio log output of the "upload to bunsen"
step as https://builder.sourceware.org/testrun/<hashid>
Cheers,
Mark
Hi Indu,
On Fri, 2022-12-23 at 07:24 -0800, Indu Bhagat wrote:
> heh, binutils-try buildbot's enthusiastic user I am, yes. I did
> wonder
> if there are enough resources for users to be doing this so frequently
> (with almost every patch set before commit), I hope its not a problem ?
It isn't a problem, a binutils build/check takes just a couple of
minutes and at least for the x86_64 and i386 builders there are more
than one worker. The most resource constrained one is the s390x worker,
but it is still able to catch up. Worst case you have to wait an hour
till your job is finally scheduled.
You can always check whether there are a lot of pending jobs at
https://builder.sourceware.org/buildbot/#/pendingbuildrequests
(There are often a couple which take multiple hours but are only ran a
few times a day)
Cheers,
Mark
From b58f0c5e82062f9c1160b72bba62a425c993fc79 Mon Sep 17 00:00:00 2001
From: Mark Wielaard <mark@klomp.org>
Date: Fri, 23 Dec 2022 14:44:53 +0100
Subject: [PATCH] Add libsframe to binutils_files
Because gdb and binutils share a git repo we only trigger builds
if they are part of gdb or binutils. libsframe is a new subdir
part of binutils.
Also add an explicit check-libsframe to binutils_step_checks.
---
builder/master.cfg | 13 ++++++++++---
1 file changed, 10 insertions(+), 3 deletions(-)
@@ -671,6 +671,7 @@ c['schedulers'].append(gccrust_bootstrap_scheduler)
binutils_files = ["bfd/",
"binutils/", "gas/", "ld/",
"libctf/",
+ "libsframe/",
"gold/", "elfcpp/",
"include/", "libiberty/", "opcodes/",
"configure", "Makefile.in"]
@@ -2106,14 +2107,17 @@ binutils_step_check = steps.Test(
workdir='binutils-build',
command=['make',
util.Interpolate('-j%(prop:ncpus)s'),
- 'check-ld', 'check-gas', 'check-binutils'],
+ 'check-ld', 'check-gas', 'check-binutils',
+ 'check-libsframe'],
name='make check',
logfiles={ "ld.sum": "ld/ld.sum",
"ld.log": "ld/ld.log",
"gas.sum": "gas/testsuite/gas.sum",
"gas.log": "gas/testsuite/gas.log",
"binutils.sum": "binutils/binutils.sum",
- "binutils.log": "binutils/binutils.log" },
+ "binutils.log": "binutils/binutils.log",
+ "libsframe.sum": "libsframe/libsframe.sum",
+ "libsframe.log": "libsframe/libsframe.log" },
haltOnFailure=False, flunkOnFailure=True)
binutils_steps_bunsen = bunsen_logfile_upload_cpio_steps(
["*.sum", "*.log"],
@@ -2123,7 +2127,8 @@ binutils_step_check_libctf = steps.Test(
workdir='binutils-build',
command=['make',
util.Interpolate('-j%(prop:ncpus)s'),
- 'check-ld', 'check-gas', 'check-binutils', 'check-libctf'],
+ 'check-ld', 'check-gas', 'check-binutils', 'check-libctf',
+ 'check-libsframe'],
name='make check',
logfiles={ "ld.sum": "ld/ld.sum",
"ld.log": "ld/ld.log",
@@ -2131,6 +2136,8 @@ binutils_step_check_libctf = steps.Test(
"gas.log": "gas/testsuite/gas.log",
"binutils.sum": "binutils/binutils.sum",
"binutils.log": "binutils/binutils.log",
+ "libsframe.sum": "libsframe/libsframe.sum",
+ "libsframe.log": "libsframe/libsframe.log",
"libctf.sum": "libctf/libctf.sum",
"libctf.log": "libctf/libctf.log" },
haltOnFailure=False, flunkOnFailure=True)
--
2.18.4