Re: [PATCH pstore-next v2 2/4] pstore: Allocate compression during late_initcall()

[Kees Cook]

On Mon, May 6, 2019 at 8:16 AM Doug Anderson <dianders@xxxxxxxxxxxx> wrote:
>
> On Sun, May 5, 2019 at 6:16 AM Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> >
> > On Fri, May 03, 2019 at 11:37:51AM -0700, Douglas Anderson wrote:
> > >
> > > On Thu, Oct 18, 2018 at 11:56 AM Kees Cook <keescook@xxxxxxxxxxxx> wrote:
> > > >
> > > > From: "Joel Fernandes (Google)" <joel@xxxxxxxxxxxxxxxxx>
> > > >
> > > > ramoops's call of pstore_register() was recently moved to run during
> > > > late_initcall() because the crypto backend may not have been ready during
> > > > postcore_initcall(). This meant early-boot crash dumps were not getting
> > > > caught by pstore any more.
> > > >
> > > > Instead, lets allow calls to pstore_register() earlier, and once crypto
> > > > is ready we can initialize the compression.
> > > >
> > > > Reported-by: Sai Prakash Ranjan <saiprakash.ranjan@xxxxxxxxxxxxxx>
> > > > Signed-off-by: Joel Fernandes (Google) <joel@xxxxxxxxxxxxxxxxx>
> > > > Tested-by: Sai Prakash Ranjan <saiprakash.ranjan@xxxxxxxxxxxxxx>
> > > > Fixes: cb3bee0369bc ("pstore: Use crypto compress API")
> > > > [kees: trivial rebase]
> > > > Signed-off-by: Kees Cook <keescook@xxxxxxxxxxxx>
> > > > ---
> > > >  fs/pstore/platform.c | 10 +++++++++-
> > > >  fs/pstore/ram.c      |  2 +-
> > > >  2 files changed, 10 insertions(+), 2 deletions(-)
> > >
> > > I'd propose that these three patches:
> > >
> > > 95047b0519c1 pstore: Refactor compression initialization
> > > 416031653eb5 pstore: Allocate compression during late_initcall()
> > > cb095afd4476 pstore: Centralize init/exit routines
> > >
> > > Get sent to linux-stable.  Specifically I'll mention that 4.19 needs
> > > it.  IMO the regression of pstore not catching early boot crashes is
> > > pretty serious IMO.
> >
> > So just those 3 commits and not this specific patch from Joel?
>
> The middle commit ("pstore: Allocate compression during
> late_initcall()") is ${SUBJECT} patch and the one with the "Fixes"
> tag.
>
> The first commit ("pstore: Centralize init/exit routines") is needed
> to apply the middle commit.
>
> I haven't done lots of analysis but the last commit ("pstore: Refactor
> compression initialization") sure looks like it's important if you
> have the middle commit.  Specifically the middle commit allocates the
> compression earlier and the last commit says that it improves handling
> of this situation.
>
>
> Unless someone thinks otherwise, it seems best to apply all 3?

At first glance, yes, but let me double-check it this morning. There
were some subtle changes that made me not initially want to backport
these, but it should be doable. :)

-- 
Kees Cook