linux-ext4 - Re: [PATCH] ext4: make ext4_mb_initialize

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

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 28 Oct 2022 16:24:52 +0530
From:   Ojaswin Mujoo <ojaswin@...ux.ibm.com>
To:     Guoqing Jiang <guoqing.jiang@...ux.dev>
Cc:     Jason Yan <yanaijie@...wei.com>, tytso@....edu,
        adilger.kernel@...ger.ca, linux-ext4@...r.kernel.org
Subject: Re: [PATCH] ext4: make ext4_mb_initialize_context return void

On Thu, Oct 27, 2022 at 04:12:45PM +0800, Guoqing Jiang wrote:
> 
> 
> On 10/27/22 2:29 PM, Jason Yan wrote:
> > 
> > On 2022/10/27 11:24, Guoqing Jiang wrote:
> > > Change the return type to void since it always return 0, and no need
> > > to do the checking in ext4_mb_new_blocks.
> > > 
> > > Signed-off-by: Guoqing Jiang <guoqing.jiang@...ux.dev>
> > > ---
> > >   fs/ext4/mballoc.c | 10 ++--------
> > >   1 file changed, 2 insertions(+), 8 deletions(-)
> > > 
> > > diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
> > > index 9dad93059945..5b2ae37a8b80 100644
> > > --- a/fs/ext4/mballoc.c
> > > +++ b/fs/ext4/mballoc.c
> > > @@ -5204,7 +5204,7 @@ static void ext4_mb_group_or_file(struct
> > > ext4_allocation_context *ac)
> > >       mutex_lock(&ac->ac_lg->lg_mutex);
> > >   }
> > >   -static noinline_for_stack int
> > > +static noinline_for_stack void
> > >   ext4_mb_initialize_context(struct ext4_allocation_context *ac,
> > >                   struct ext4_allocation_request *ar)
> > >   {
> > > @@ -5253,8 +5253,6 @@ ext4_mb_initialize_context(struct
> > > ext4_allocation_context *ac,
> > >               (unsigned) ar->lleft, (unsigned) ar->pleft,
> > >               (unsigned) ar->lright, (unsigned) ar->pright,
> > >               inode_is_open_for_write(ar->inode) ? "" : "non-");
> > > -    return 0;
> > > -
> > >   }
> > >     static noinline_for_stack void
> > > @@ -5591,11 +5589,7 @@ ext4_fsblk_t ext4_mb_new_blocks(handle_t *handle,
> > >           goto out;
> > >       }
> > >   -    *errp = ext4_mb_initialize_context(ac, ar);
> > > -    if (*errp) {
> > > -        ar->len = 0;
> > > -        goto out;
> > > -    }
> > > +    ext4_mb_initialize_context(ac, ar);
> > 
> > This changed the logic here slightly. *errp will not be intialized with
> > zero after this change. So we need to carefully check whether this will
> > cause any issues.
> 
> Yes, thanks for reminder. I think "*errp" is always set later with below.
> 
> https://elixir.bootlin.com/linux/v6.1-rc2/source/fs/ext4/mballoc.c#L5606
> https://elixir.bootlin.com/linux/v6.1-rc2/source/fs/ext4/mballoc.c#L5611
> https://elixir.bootlin.com/linux/v6.1-rc2/source/fs/ext4/mballoc.c#L5629
> https://elixir.bootlin.com/linux/v6.1-rc2/source/fs/ext4/mballoc.c#L5646
Hi Guoqing,

I agree, it seems to be intialized correctly later in the code. The
flow is something like:

  ext4_fsblk_t ext4_mb_new_blocks(...)
  {
      ...
      ext4_mb_initialize_context(ac, ar);
      ...
      if (!ext4_mb_use_preallocated(ac)) {
          *errp = ext4_mb_pa_alloc(ac);  // *errp init to 0 on success
          ...
      }

      if (likely(ac->ac_status == AC_STATUS_FOUND)) {
          // *errp init to 0 on success
          *errp = ext4_mb_mark_diskspace_used(ac, handle, reserv_clstrs);
          ...
      } else {
          ...
          *errp = -ENOSPC;
      }
      ...
  }

So it seems like this cleanup won't alter the behavior. Feel free to
add:

Reviewed-by: Ojaswin Mujoo <ojaswin@...ux.ibm.com>

Regards,
ojaswin