From: "Bob Feng" <bob.c.feng@intel.com>
To: Laszlo Ersek <lersek@redhat.com>,
"devel@edk2.groups.io" <devel@edk2.groups.io>
Cc: "Gao, Liming" <liming.gao@intel.com>
Subject: Re: [edk2-devel] [Patch 10/10 V3] BaseTools: Enable block queue log agent.
Date: Tue, 23 Jul 2019 12:00:57 +0000 [thread overview]
Message-ID: <08650203BA1BD64D8AD9B6D5D74A85D160B468EB@SHSMSX105.ccr.corp.intel.com> (raw)
In-Reply-To: <e3f68d77-4837-0ef6-ab4f-95e50c4621ff@redhat.com>
Good suggestion. I think ThreadNum * 10 is better than a const. I'll update it in patches V4.
I'd like to collect more comments on other parts and update all the comments in V4.
Thanks,
Bob
-----Original Message-----
From: Laszlo Ersek [mailto:lersek@redhat.com]
Sent: Tuesday, July 23, 2019 6:12 PM
To: devel@edk2.groups.io; Feng, Bob C <bob.c.feng@intel.com>
Cc: Gao, Liming <liming.gao@intel.com>
Subject: Re: [edk2-devel] [Patch 10/10 V3] BaseTools: Enable block queue log agent.
On 07/23/19 05:58, Bob Feng wrote:
> BZ: https://bugzilla.tianocore.org/show_bug.cgi?id=1875
>
> To support Ctrl+S and Ctrl+Q, we enable block queue for log.
>
> Cc: Liming Gao <liming.gao@intel.com>
> Signed-off-by: Bob Feng <bob.c.feng@intel.com>
> ---
> BaseTools/Source/Python/Common/EdkLogger.py | 10 ++++++----
> BaseTools/Source/Python/build/build.py | 8 +++++---
> 2 files changed, 11 insertions(+), 7 deletions(-)
>
> diff --git a/BaseTools/Source/Python/Common/EdkLogger.py
> b/BaseTools/Source/Python/Common/EdkLogger.py
> index 15fd1458a95a..06da4a9d0a1d 100644
> --- a/BaseTools/Source/Python/Common/EdkLogger.py
> +++ b/BaseTools/Source/Python/Common/EdkLogger.py
> @@ -93,11 +93,13 @@ except:
> """
> try:
> self.enqueue(self.prepare(record))
> except Exception:
> self.handleError(record)
> -
> +class BlockQueueHandler(QueueHandler):
> + def enqueue(self, record):
> + self.queue.put(record,True)
> ## Log level constants
> DEBUG_0 = 1
> DEBUG_1 = 2
> DEBUG_2 = 3
> DEBUG_3 = 4
> @@ -290,23 +292,23 @@ def LogClientInitialize(log_q):
> # Since we use different format to log different levels of message into different
> # place (stdout or stderr), we have to use different "Logger" objects to do this.
> #
> # For DEBUG level (All DEBUG_0~9 are applicable)
> _DebugLogger.setLevel(INFO)
> - _DebugChannel = QueueHandler(log_q)
> + _DebugChannel = BlockQueueHandler(log_q)
> _DebugChannel.setFormatter(_DebugFormatter)
> _DebugLogger.addHandler(_DebugChannel)
>
> # For VERBOSE, INFO, WARN level
> _InfoLogger.setLevel(INFO)
> - _InfoChannel = QueueHandler(log_q)
> + _InfoChannel = BlockQueueHandler(log_q)
> _InfoChannel.setFormatter(_InfoFormatter)
> _InfoLogger.addHandler(_InfoChannel)
>
> # For ERROR level
> _ErrorLogger.setLevel(INFO)
> - _ErrorCh = QueueHandler(log_q)
> + _ErrorCh = BlockQueueHandler(log_q)
> _ErrorCh.setFormatter(_ErrorFormatter)
> _ErrorLogger.addHandler(_ErrorCh)
>
> ## Set log level
> #
> diff --git a/BaseTools/Source/Python/build/build.py
> b/BaseTools/Source/Python/build/build.py
> index 4125b2832946..603d3aa6dad4 100644
> --- a/BaseTools/Source/Python/build/build.py
> +++ b/BaseTools/Source/Python/build/build.py
> @@ -2046,14 +2046,15 @@ class Build():
> for m in Pa.GetAllModuleInfo:
> mqueue.put(m)
> data_pipe_file = os.path.join(Pa.BuildDir, "GlobalVar_%s_%s.bin" % (str(Pa.Guid),Pa.Arch))
> Pa.DataPipe.dump(data_pipe_file)
> autogen_rt = self.StartAutoGen(mqueue, Pa.DataPipe, self.SkipAutoGen, PcdMaList,self.share_data)
> - self.Progress.Stop("done!")
> - self.AutoGenTime += int(round((time.time() - AutoGenStart)))
> +
> if not autogen_rt:
> return
> + self.AutoGenTime += int(round((time.time() - AutoGenStart)))
> + self.Progress.Stop("done!")
> for Arch in Wa.ArchList:
> MakeStart = time.time()
> for Ma in BuildModules:
> # Generate build task for the module
> if not Ma.IsBinaryModule:
> @@ -2294,17 +2295,18 @@ def LogBuildTime(Time):
> # if it's executed successfully or not.
> #
> # @retval 0 Tool was successful
> # @retval 1 Tool failed
> #
> +LogQMaxSize = 60
Can we use a value here that is proportional to multiprocessing.cpu_count()?
Or, more precisely, proportional to the thread count set with "-n"?
Like, multiply the thread count by 10.
Because, in case the thread count is really high, a constant log queue size could throttle compilation, even if terminal output is not suspended.
The factor 10 is still quite arbitrary, but I believe it should block compilation "soon enough" after terminal output is suspended, and should not throttle compilation in practice when terminal output is not suspended.
Just an idea.
Thanks
Laszlo
> def Main():
> StartTime = time.time()
>
> #
> # Create a log Queue
> #
> - LogQ = mp.Queue()
> + LogQ = mp.Queue(LogQMaxSize)
> # Initialize log system
> EdkLogger.LogClientInitialize(LogQ)
> GlobalData.gCommand = sys.argv[1:]
> #
> # Parse the options and args
>
next prev parent reply other threads:[~2019-07-23 12:01 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-07-23 3:58 [Patch 00/10 V3] Enable multiple process AutoGen Bob Feng
2019-07-23 3:58 ` [Patch 01/10 V3] BaseTools: Singleton the object to handle build conf file Bob Feng
2019-07-23 3:58 ` [Patch 02/10 V3] BaseTools: Split WorkspaceAutoGen._InitWorker into multiple functions Bob Feng
2019-07-23 3:58 ` [Patch 03/10 V3] BaseTools: Add functions to get platform scope build options Bob Feng
2019-07-23 3:58 ` [Patch 04/10 V3] BaseTools: Decouple AutoGen Objects Bob Feng
2019-07-23 3:58 ` [Patch 05/10 V3] BaseTools: Enable Multiple Process AutoGen Bob Feng
2019-07-23 3:58 ` [Patch 06/10 V3] BaseTools: Add shared data for processes Bob Feng
2019-07-23 3:58 ` [Patch 07/10 V3] BaseTools: Add LogAgent to support multiple process Autogen Bob Feng
2019-07-23 3:58 ` [Patch 08/10 V3] BaseTools: Move BuildOption parser out of build.py Bob Feng
2019-07-23 3:58 ` [Patch 09/10 V3] BaseTools: Add the support for python 2 Bob Feng
2019-07-23 3:58 ` [Patch 10/10 V3] BaseTools: Enable block queue log agent Bob Feng
2019-07-23 10:11 ` [edk2-devel] " Laszlo Ersek
2019-07-23 12:00 ` Bob Feng [this message]
2019-07-23 11:30 ` [edk2-devel] [Patch 00/10 V3] Enable multiple process AutoGen Laszlo Ersek
2019-07-30 13:38 ` Bob Feng
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-list 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=08650203BA1BD64D8AD9B6D5D74A85D160B468EB@SHSMSX105.ccr.corp.intel.com \
--to=devel@edk2.groups.io \
/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