Openembedded Core Discussions
 help / color / mirror / Atom feed
From: Richard Purdie <richard.purdie@linuxfoundation.org>
To: Mike Crowe <mac@mcrowe.com>
Cc: openembedded-core@lists.openembedded.org
Subject: Re: [PATCH 1/2] image.py: avoid mkdir race when building multiple images
Date: Thu, 19 Nov 2015 11:23:47 +0000	[thread overview]
Message-ID: <1447932227.12500.121.camel@linuxfoundation.org> (raw)
In-Reply-To: <1447932037-29676-1-git-send-email-mac@mcrowe.com>

On Thu, 2015-11-19 at 11:20 +0000, Mike Crowe wrote:
> If multiple images are being built simultaneously against the same
> sysroot then the call to os.makedirs in Image._write_wic_env can fail
> with:
> 
> File: '.../meta/lib/oe/image.py', lineno: 341, function: _write_wic_env
>      0337:        """
>      0338:        stdir = self.d.getVar('STAGING_DIR_TARGET', True)
>      0339:        outdir = os.path.join(stdir, 'imgdata')
>      0340:        if not os.path.exists(outdir):
>  *** 0341:            os.makedirs(outdir)
>      0342:        basename = self.d.getVar('IMAGE_BASENAME', True)
>      0343:        with open(os.path.join(outdir, basename) + '.env', 'w') as envf:
>      0344:            for var in self.d.getVar('WICVARS', True).split():
>      0345:                value = self.d.getVar(var, True)
> File: '/usr/lib/python2.7/os.py', lineno: 157, function: makedirs
>      0153:            if e.errno != errno.EEXIST:
>      0154:                raise
>      0155:        if tail == curdir:           # xxx/newdir/. exists if xxx/newdir exists
>      0156:            return
>  *** 0157:    mkdir(name, mode)
>      0158:
>      0159:def removedirs(name):
>      0160:    """removedirs(path)
>      0161:
> Exception: OSError: [Errno 17] File exists: '.../tmp-glibc/sysroots/cheetah/imgdata'
> 
> os.makedirs protects against racing on parent directories but not on the
> final part.
> 
> The sysroot directory will already exist so let's just try and make the
> imgdata directory and cope with the exception thrown if it already
> exists. This stops the race being fatal.
> 
> Once the directory has been created, there's no race on the actual file
> since the filename contains IMAGE_BASENAME.
> 
> Signed-off-by: Mike Crowe <mac@mcrowe.com>
> ---
>  meta/lib/oe/image.py | 13 +++++++++++--
>  1 file changed, 11 insertions(+), 2 deletions(-)
> 
> diff --git a/meta/lib/oe/image.py b/meta/lib/oe/image.py
> index b9eb3de..8c797ec 100644
> --- a/meta/lib/oe/image.py
> +++ b/meta/lib/oe/image.py
> @@ -1,5 +1,6 @@
>  from oe.utils import execute_pre_post_process
>  import os
> +import errno
>  import subprocess
>  import multiprocessing
>  
> @@ -337,8 +338,16 @@ class Image(ImageDepGraph):
>          """
>          stdir = self.d.getVar('STAGING_DIR_TARGET', True)
>          outdir = os.path.join(stdir, 'imgdata')
> -        if not os.path.exists(outdir):
> -            os.makedirs(outdir)
> +        try:
> +            # This may race against another image being created, so
> +            # let's just try and create the directory and just ignore
> +            # the error if it already exists. The parent directory
> +            # will already exist so there's no benefit to using
> +            # os.makedirs (which is also subject to the race.)
> +            os.mkdir(outdir)
> +        except OSError as e:
> +            if e.errno != errno.EEXIST:
> +                raise
>          basename = self.d.getVar('IMAGE_BASENAME', True)
>          with open(os.path.join(outdir, basename) + '.env', 'w') as envf:
>              for var in self.d.getVar('WICVARS', True).split():

How about using bb.utils.mkdirhier() which already does this?

Cheers,

Richard




  reply	other threads:[~2015-11-19 11:24 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-11-19 11:20 [PATCH 1/2] image.py: avoid mkdir race when building multiple images Mike Crowe
2015-11-19 11:23 ` Richard Purdie [this message]
2015-11-19 11:50   ` Mike Crowe

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-all 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=1447932227.12500.121.camel@linuxfoundation.org \
    --to=richard.purdie@linuxfoundation.org \
    --cc=mac@mcrowe.com \
    --cc=openembedded-core@lists.openembedded.org \
    /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