From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mail-lj1-f195.google.com (mail-lj1-f195.google.com [209.85.208.195]) by mx.groups.io with SMTP id smtpd.web11.2559.1586287191921342875 for ; Tue, 07 Apr 2020 12:19:52 -0700 Authentication-Results: mx.groups.io; dkim=missing; spf=pass (domain: gmail.com, ip: 209.85.208.195, mailfrom: ricardo.ribalda@gmail.com) Received: by mail-lj1-f195.google.com with SMTP id z26so906381ljz.11 for ; Tue, 07 Apr 2020 12:19:51 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:mime-version:references:in-reply-to:from:date :message-id:subject:to:cc; bh=EdMJ7/BYYXKPGrK60sXBk2GquyWoi6KLsZD3uc8n6JE=; b=MLvLb+/YlNnjI/lk1e79kdoBj+nWFi1zBrbtUrosPWKFUzKGMChE8xqlCw0UEqVxqP XSeQnijpjngK5XJDWIVsZzRWQTjCO7kDY86GYPV9drWh2yWwFHslRAI2f0mmV4DRmyx/ u3xi9rr9LGnT8j8huYuZLYdFe3jLgHZWGwtMqRNEr49lj+b37lki7wBM9/nrbhS7aRdN kje+rOkQLUDfnKxkDZVvJXr3TXXyAGNgvj4w6zP+QRssngOl///23uQbadc0BK9RdJ43 mc/BGuj9lvh5fRdM06wxcN5Mk9gJE/ksXqLdPj/0qzF3t4csQcJBDE0sLhNsBctjoxkw Njjw== X-Gm-Message-State: AGi0PuYG6ilUGOtnRdLLYXzldObjWk1pQh/AxXhcS4fVWr+mP42lhuEC exTVZ4EsCHtqlN/OFpOKfBvlqYaXsDc7byYAd20= X-Google-Smtp-Source: APiQypKmEuynmpqhcI41d9K0bHmFkV66OpW/9XKUjUJskWpHpqm1YdBc2ZIIphq94AKTZtqfydG1qTV60ar5aAshc+Q= X-Received: by 2002:a2e:8746:: with SMTP id q6mr2595549ljj.13.1586287189956; Tue, 07 Apr 2020 12:19:49 -0700 (PDT) MIME-Version: 1.0 References: <20200304083438.1022216-1-ricardo@ribalda.com> <20200304095334.1f20ddd9@ub1910> <20200305092855.1f9ccae8@ub1910> <20200407191256.6cb45445@ub1910> <20200407200236.748931eb@ub1910> In-Reply-To: <20200407200236.748931eb@ub1910> From: "Ricardo Ribalda" Date: Tue, 7 Apr 2020 21:19:32 +0200 Message-ID: Subject: Re: [OE-core] [PATCH 1/2] wic: Fix permissions when using exclude or include path To: Paul Barker Cc: Richard Purdie , openembedded-core Content-Type: text/plain; charset="UTF-8" Hi On Tue, Apr 7, 2020 at 9:02 PM Paul Barker wrote: > > On Tue, 7 Apr 2020 20:40:18 +0200 > Ricardo Ribalda Delgado wrote: > > > Hi Paul > > > > Thanks for your review, It has been already merged, but if there is > > something wrong we can send a patch fixing it. > > > > https://git.openembedded.org/openembedded-core/commit/?id=36993eea89d1c011397b7692b9b8d61b499d0171 > > > > On Tue, Apr 7, 2020 at 8:13 PM Paul Barker wrote: > > > > > > On Fri, 3 Apr 2020 21:53:39 +0200 > > > Ricardo Ribalda Delgado wrote: > > > > > > > ping? > > > > > > I think that '../pseudo' should not be used here. I'll explain inline... > > > > > > > > > > > > > This results in a rootfs owned by the user that is running the wic > > > > > command (usually UID 1000), which makes some rootfs unbootable. > > > > > > > > > > To fix this we copy the content of the pseudo folders to the new folder > > > > > and modify the pseudo database using the "pseudo -B" command. > > > > > > > > > > Signed-off-by: Ricardo Ribalda Delgado > > > > > --- > > > > > scripts/lib/wic/plugins/source/rootfs.py | 22 +++++++++++++++++++--- > > > > > 1 file changed, 19 insertions(+), 3 deletions(-) > > > > > > > > > > diff --git a/scripts/lib/wic/plugins/source/rootfs.py b/scripts/lib/wic/plugins/source/rootfs.py > > > > > index 705aeb5563..40419a64b3 100644 > > > > > --- a/scripts/lib/wic/plugins/source/rootfs.py > > > > > +++ b/scripts/lib/wic/plugins/source/rootfs.py > > > > > @@ -16,11 +16,11 @@ import os > > > > > import shutil > > > > > import sys > > > > > > > > > > -from oe.path import copyhardlinktree > > > > > +from oe.path import copyhardlinktree, copytree > > > > > > > > > > from wic import WicError > > > > > from wic.pluginbase import SourcePlugin > > > > > -from wic.misc import get_bitbake_var > > > > > +from wic.misc import get_bitbake_var, exec_native_cmd > > > > > > > > > > logger = logging.getLogger('wic') > > > > > > > > > > @@ -44,6 +44,15 @@ class RootfsPlugin(SourcePlugin): > > > > > > > > > > return os.path.realpath(image_rootfs_dir) > > > > > > > > > > + @staticmethod > > > > > + def __get_pseudo(native_sysroot, rootfs): > > > > > + pseudo = "export PSEUDO_PREFIX=%s/usr;" % native_sysroot > > > > > + pseudo += "export PSEUDO_LOCALSTATEDIR=%s;" % os.path.join(rootfs, "../pseudo") > > > > > + pseudo += "export PSEUDO_PASSWD=%s;" % rootfs > > > > > + pseudo += "export PSEUDO_NOSYMLINKEXP=1;" > > > > > + pseudo += "%s " % get_bitbake_var("FAKEROOTCMD") > > > > > + return pseudo > > > > > + > > > > > @classmethod > > > > > def do_prepare_partition(cls, part, source_params, cr, cr_workdir, > > > > > oe_builddir, bootimg_dir, kernel_dir, > > > > > @@ -78,9 +87,16 @@ class RootfsPlugin(SourcePlugin): > > > > > > > > > > if os.path.lexists(new_rootfs): > > > > > shutil.rmtree(os.path.join(new_rootfs)) > > > > > - > > > > > copyhardlinktree(part.rootfs_dir, new_rootfs) > > > > > > > > > > + if os.path.lexists(os.path.join(new_rootfs, "../pseudo")): > > > > > > new_rootfs is set by the following statement a few lines above: > > > > > > new_rootfs = os.path.realpath(os.path.join(cr_workdir, "rootfs%d" % part.lineno)) > > > > > > Consider that `cr_workdir` may contain multiple rootfs staging directories > > > corresponding to multiple lines in the wks file, for example if a rootfs > > > image is duplicated into multiple partitions for redundancy. In that case > > > `os.path.join(new_rootfs, "../pseudo")` will clash between these different > > > rootfs copies. > > > > > > Let's use an explicit path instead, such as: > > > > > > new_pseudo = os.path.realpath(os.path.join(cr_workdir, "pseudo%d" % part.lineno)) > > > > The reason to have that path was to follow the same structure as the > > real image.bb. > > > > If there are multiple partitions on the .wic file the different > > partitions are done one by one, not > > in parallel. > > > > So > > ../pseudo will be created for partition1 > > then it will be used to generate the partition1 > > > > ../pseudo will be deleted > > ../pseudo will be created for partition2 > > > > Even if they use the same partition, the code works (and ../pseudo is > > useless once the partition is generated) > > > > Having these separate is important for debugging though, it lets you look > through the different copies after wic exits if something is wrong. I see your point, I can make a patch for that > > > > > > > > > + shutil.rmtree(os.path.join(new_rootfs, "../pseudo")) > > > > > + copytree(os.path.join(part.rootfs_dir, "../pseudo"), > > > > > > part.rootfs_dir is whatever is given as the option to `--rootfs-dir`. There > > > is no guarantee that `../psuedo` is valid or if it corresponds to the rootfs > > > directory given. It's unsafe to step up the directory tree and make > > > assumptions like this. > > > > I think that if we do not pass a real rootfs to the rootfs plugin it > > is an error from the user. > > > > We can add a more beautiful error message instead of a backtrace. > > > > Or if you believe that it is a valid usecase to not pass a rootfs then > > we can continue with a warning/debug message and explicitly telling > > the user that the permissions are going to be invalid, because what he > > is using as a roofs is an unknow directory for bitbake. > > This is a valid and existing usecase. This is how data partitions are > populated and how you separate /home or another directory into its own > partiton (e.g. > https://stackoverflow.com/questions/56187209/yocto-create-and-populate-a-separate-home-partition). Luckily on the usecase on that stackoverflow the patch will still work. we need to combine partition split + embed-roots/include_path/exlude_path to make it fail By the way, by looking at your example I think that all the partition split executed from outside bitbake will have invalid partitions on their content. We need to make something like: part /home --source rootfs --root-dir=/home --ondisk sda --fstype=ext4 --label home > > > > > I have no personal preference for any of the two, tell me what do you > > prefer (or a different option) and I will implement it. > > > > Thans again for the review. > > > > This patch needs reverting from master/dunfell. I hope it hasn't gone into > the M4 build... Instead of reverting I believe it is simpler to prepare a patch that solves both of your concerns - pseduoX instead of ../pseudo (although this does not break the code) - keep the old behaviour if the pseudo folder is not found (wrong permissions, but no backtrace) And we can start talking about a way to split partitions and keep the right permissions/username Best regards. > > -- > Paul Barker > Konsulko Group > -- Ricardo Ribalda