[PATCH RESEND] erofs-utils: fix a memory leak of multiple devices

[Yue Hu]

On Fri, 22 Jul 2022 11:28:27 +0800
Gao Xiang <hsiangkao at linux.alibaba.com> wrote:

> Hi Yue,
> 
> On Fri, Jul 22, 2022 at 11:10:08AM +0800, Yue Hu wrote:
> > The memory allocated for multiple devices should be freed when to exit.
> > Let's add a helper to fix it since there is more than one to use it.
> > 
> > Signed-off-by: Yue Hu <huyue2 at coolpad.com>
> > ---
> >  dump/main.c              | 7 ++++---
> >  fsck/main.c              | 7 ++++---
> >  fuse/main.c              | 5 +++--
> >  include/erofs/internal.h | 1 +
> >  lib/super.c              | 6 ++++++
> >  5 files changed, 18 insertions(+), 8 deletions(-)
> > 
> > diff --git a/dump/main.c b/dump/main.c
> > index 40e850a..c9b3a8f 100644
> > --- a/dump/main.c
> > +++ b/dump/main.c
> > @@ -615,7 +615,7 @@ int main(int argc, char **argv)
> >  	err = erofs_read_superblock();
> >  	if (err) {
> >  		erofs_err("failed to read superblock");
> > -		goto exit_dev_close;
> > +		goto exit_put_super;
> >  	}
> >  
> >  	if (!dumpcfg.totalshow) {
> > @@ -630,13 +630,14 @@ int main(int argc, char **argv)
> >  
> >  	if (dumpcfg.show_extent && !dumpcfg.show_inode) {
> >  		usage();
> > -		goto exit_dev_close;
> > +		goto exit_put_super;
> >  	}
> >  
> >  	if (dumpcfg.show_inode)
> >  		erofsdump_show_fileinfo(dumpcfg.show_extent);
> >  
> > -exit_dev_close:
> > +exit_put_super:
> > +	erofs_put_super();
> >  	dev_close();
> >  exit:
> >  	blob_closeall();
> > diff --git a/fsck/main.c b/fsck/main.c
> > index 5a2f659..a8f0e24 100644
> > --- a/fsck/main.c
> > +++ b/fsck/main.c
> > @@ -813,12 +813,12 @@ int main(int argc, char **argv)
> >  	err = erofs_read_superblock();
> >  	if (err) {
> >  		erofs_err("failed to read superblock");
> > -		goto exit_dev_close;
> > +		goto exit_put_super;  
> 
> Why do we call erofs_put_super() again here? I think we don't need to
> call erofs_put_super for all failed paths.

There is a call to dev_read() which may fails after allocating memory.
Let me send v2 for this.

> 
> Thanks,
> Gao Xiang