Re: [PATCH] net/rds: Avoid potential use after free in rds_send_remove_from_sock

From: Leon Romanovsky
Date: Wed Apr 21 2021 - 11:36:40 EST


On Wed, Apr 21, 2021 at 04:19:03PM +0200, Krzysztof Kozlowski wrote:
> On Tue, 20 Apr 2021 at 11:13, Leon Romanovsky <leon@xxxxxxxxxx> wrote:
> >
> > On Tue, Apr 20, 2021 at 12:09:06PM +0300, Leon Romanovsky wrote:
> > > On Tue, Apr 06, 2021 at 07:09:12PM -0500, Aditya Pakki wrote:
> > > > In case of rs failure in rds_send_remove_from_sock(), the 'rm' resource
> > > > is freed and later under spinlock, causing potential use-after-free.
> > > > Set the free pointer to NULL to avoid undefined behavior.
> > > >
> > > > Signed-off-by: Aditya Pakki <pakki001@xxxxxxx>
> > > > ---
> > > > net/rds/message.c | 1 +
> > > > net/rds/send.c | 2 +-
> > > > 2 files changed, 2 insertions(+), 1 deletion(-)
> > >
> > > Dave, Jakub
> > >
> > > Please revert this patch, given responses from Eric and Al together
> > > with this response from Greg here https://lore.kernel.org/lkml/YH5/i7OvsjSmqADv@xxxxxxxxx
> >
> > https://lore.kernel.org/lkml/YH5%2Fi7OvsjSmqADv@xxxxxxxxx/
> >
> > >
> > > BTW, I looked on the rds code too and agree with Eric, this patch
> > > is a total garbage.
>
> When reverting, consider giving credits to Kees/Coverity as he pointed
> out after testing linux-next that this is bogus:
> https://lore.kernel.org/linux-next/202104081640.1A09A99900@keescook/

The revert is already done for almost all @umn.edu patches.
https://lore.kernel.org/lkml/20210421130105.1226686-1-gregkh@xxxxxxxxxxxxxxxxxxx/

Thanks

>
>
> Best regards,
> Krzysztof