| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <20181012054923.rncvdtkqiau3pdd5@localhost.localdomain>
Date: Fri, 12 Oct 2018 07:49:23 +0200
From: Lukas Czerner <lczerner@...hat.com>
To: "Darrick J. Wong" <darrick.wong@...cle.com>
Cc: linux-ext4@...r.kernel.org
Subject: Re: [PATCH] fuse2fs: return proper exit code from fuse_main
On Thu, Oct 11, 2018 at 08:28:40AM -0700, Darrick J. Wong wrote:
> On Thu, Oct 11, 2018 at 03:43:36PM +0200, Lukas Czerner wrote:
> > Currently fuse2fs will always return 0 exit code indicating successful
> > operation even though the mount failed either because we failed to
> > properly read the file system in the first place, or the fuse_main()
> > failed for whatever reason.
> >
> > Fix it by using the return code from fuse_main(), or return 32 in case
> > we fail because the file system is corrupted, or we encountered a
> > problem preventing us mounting the file system. 32 because this is a
> > libmount exit code indicating mount failed.
> >
> > Signed-off-by: Lukas Czerner <lczerner@...hat.com>
> > ---
> > misc/fuse2fs.c | 17 +++++++++++------
> > 1 file changed, 11 insertions(+), 6 deletions(-)
> >
> > diff --git a/misc/fuse2fs.c b/misc/fuse2fs.c
> > index 5c73895e..d7a0b668 100644
> > --- a/misc/fuse2fs.c
> > +++ b/misc/fuse2fs.c
> > @@ -3720,7 +3720,7 @@ int main(int argc, char *argv[])
> > {
> > struct fuse_args args = FUSE_ARGS_INIT(argc, argv);
> > struct fuse2fs fctx;
> > - errcode_t err;
> > + errcode_t err = 0;
> > char *logfile;
> > char extra_args[BUFSIZ];
> > int ret = 0, flags = EXT2_FLAG_64BITS | EXT2_FLAG_EXCLUSIVE;
> > @@ -3753,6 +3753,7 @@ int main(int argc, char *argv[])
> > fctx.err_fp = fopen(logfile, "a");
> > if (!fctx.err_fp) {
> > perror(logfile);
> > + err = errno;
> > goto out;
> > }
> > } else
> > @@ -3766,7 +3767,6 @@ int main(int argc, char *argv[])
> > }
> >
> > /* Start up the fs (while we still can use stdout) */
> > - ret = 2;
> > if (!fctx.ro)
> > flags |= EXT2_FLAG_RW;
> > err = ext2fs_open2(fctx.device, NULL, flags, 0, 0, unix_io_manager,
> > @@ -3779,8 +3779,6 @@ int main(int argc, char *argv[])
> > fctx.fs = global_fs;
> > global_fs->priv_data = &fctx;
> >
> > - ret = 3;
> > -
> > if (ext2fs_has_feature_journal_needs_recovery(global_fs->super)) {
> > if (!fctx.ro) {
> > printf(_("%s: recovering journal\n"), fctx.device);
> > @@ -3797,6 +3795,7 @@ int main(int argc, char *argv[])
> > } else {
> > printf("%s", _("Journal needs recovery; running "
> > "`e2fsck -E journal_only' is required.\n"));
> > + err = 1;
> > goto out;
> > }
> > }
> > @@ -3836,6 +3835,7 @@ int main(int argc, char *argv[])
> > if (global_fs->super->s_state & EXT2_ERROR_FS) {
> > printf("%s",
> > _("Errors detected; running e2fsck is required.\n"));
> > + err = 1;
> > goto out;
> > }
> >
> > @@ -3859,11 +3859,16 @@ int main(int argc, char *argv[])
> > }
> >
> > pthread_mutex_init(&fctx.bfl, NULL);
> > - fuse_main(args.argc, args.argv, &fs_ops, &fctx);
> > + ret = fuse_main(args.argc, args.argv, &fs_ops, &fctx);
>
> Hmmm, what /does/ fuse_main return? According to the libfuse github
> site it returns 0 for success and "nonzero" for failure. The source
> code seems to return 1 on failure (I think), but we probably ought to
> set ret to 1 ("incorrect invocation or permissions") or 32 explicitly
> just in case they ever change their minds...
Ah yeah, I was assuming (wrongly) that the fuse_main would return the
libmount codes, which is not true. The source code for libfuse says:
* The following error codes may be returned from fuse_main():
* 1: Invalid option arguments
* 2: No mount point specified
* 3: FUSE setup failed
* 4: Mounting failed
* 5: Failed to daemonize (detach from session)
* 6: Failed to set up signal handlers
* 7: An error occured during the life of the file system
so the question is, do we want to map it somehow to the libmount error
codes ?
* [u]mount(8) exit code 0: no errors
* [u]mount(8) exit code 1: incorrect invocation or permission
* [u]mount(8) exit code 2: out of memory, cannot fork, ...
* [u]mount(8) exit code 4: internal mount bug or wrong version
* [u]mount(8) exit code 8: user interrupt
* [u]mount(8) exit code 16: problems writing, locking, ... mtab/utab
* [u]mount(8) exit code 32: mount failure
* [u]mount(8) exit code 64: some mount succeeded; usually when executed with
* --all options. Never returned by libmount.
Or just return whatever we get from the main_fuse() and 1 for other
errors ? I do not really mind either way.
-Lukas
>
> > pthread_mutex_destroy(&fctx.bfl);
> >
> > - ret = 0;
> > out:
> > + /*
> > + * Encountered error reading the file system. Return standard "mount
> > + * failure" mount exit code (32).
> > + */
> > + if (err)
> > + ret = 32;
>
> ...I guess "mount failure" for libext2fs problems is good enough, though
> part of me thinks that we should return 1 if ext2fs_open can't open the
> block device due to EPERM/EACCESS.
>
> <shrug> OTOH "mount failure" is sufficiently vague to hide just about
> anything behind. :)
>
> --D
>
> > if (global_fs) {
> > err = ext2fs_close(global_fs);
> > if (err)
> > --
> > 2.17.1
> >
Powered by blists - more mailing lists