[PATCH] ath10k: Fix the parsing error in service available event

[Doug Anderson]

Hi,

On Wed, Oct 28, 2020 at 8:47 AM Rakesh Pillai <pillair at codeaurora.org> wrote:
>
> > -----Original Message-----
> > From: Doug Anderson <dianders at chromium.org>
> > Sent: Wednesday, October 28, 2020 8:07 PM
> > To: Rakesh Pillai <pillair at codeaurora.org>
> > Cc: ath10k <ath10k at lists.infradead.org>; linux-wireless <linux-
> > wireless at vger.kernel.org>; LKML <linux-kernel at vger.kernel.org>; Abhishek
> > Kumar <kuabhs at chromium.org>; Brian Norris <briannorris at chromium.org>
> > Subject: Re: [PATCH] ath10k: Fix the parsing error in service available event
> >
> > Hi,
> >
> > On Tue, Oct 27, 2020 at 8:20 AM Rakesh Pillai <pillair at codeaurora.org>
> > wrote:
> > >
> > > The wmi service available event has been
> > > extended to contain extra 128 bit for new services
> > > to be indicated by firmware.
> > >
> > > Currently the presence of any optional TLVs in
> > > the wmi service available event leads to a parsing
> > > error with the below error message:
> > > ath10k_snoc 18800000.wifi: failed to parse svc_avail tlv: -71
> > >
> > > The wmi service available event parsing should
> > > not return error for the newly added optional TLV.
> > > Fix this parsing for service available event message.
> > >
> > > Tested-on: WCN3990 hw1.0 SNOC
> > >
> > > Signed-off-by: Rakesh Pillai <pillair at codeaurora.org>
> > > ---
> > >  drivers/net/wireless/ath/ath10k/wmi-tlv.c | 3 +++
> > >  1 file changed, 3 insertions(+)
> > >
> > > diff --git a/drivers/net/wireless/ath/ath10k/wmi-tlv.c
> > b/drivers/net/wireless/ath/ath10k/wmi-tlv.c
> > > index 932266d..3b49e29 100644
> > > --- a/drivers/net/wireless/ath/ath10k/wmi-tlv.c
> > > +++ b/drivers/net/wireless/ath/ath10k/wmi-tlv.c
> > > @@ -1404,9 +1404,12 @@ static int ath10k_wmi_tlv_svc_avail_parse(struct
> > ath10k *ar, u16 tag, u16 len,
> > >                 arg->service_map_ext_len = *(__le32 *)ptr;
> > >                 arg->service_map_ext = ptr + sizeof(__le32);
> > >                 return 0;
> > > +       case WMI_TLV_TAG_FIRST_ARRAY_ENUM:
> > > +               return 0;
> >
> > This is at least slightly worrying to me.  If I were calling this
> > function, I'd expect that if I didn't get back an error that at least
> > "arg->service_map_ext_len" was filled in.  Seems like you should do:
> >
> > case WMI_TLV_TAG_FIRST_ARRAY_ENUM:
> >   arg->service_map_ext_len = 0;
> >   arg->service_map_ext = NULL;
> >   return 0;
> >
> > ...and maybe add a comment about why you're doing that?
> >
> > At the moment things are working OK because
> > ath10k_wmi_event_service_available() happens to init the structure to
> > 0 before calling with:
> >
> >   struct wmi_svc_avail_ev_arg arg = {};
> >
> > ....but it doesn't seem like a great idea to rely on that.
> >
> > That all being said, I'm just a drive-by reviewer and if everyone else
> > likes it the way it is, feel free to ignore my comments.
>
>
> Hi Doug,
>
> The TLV TAG " WMI_TLV_TAG_STRUCT_SERVICE_AVAILABLE_EVENT" is the first and a mandatory TLV in the service available event.
> The subsequent TLVs are optional ones and may or may not be present (based on FW versions).
> This patch just fixes the bug, where the presence of any other TLVs are leading to a failure in parsing the service available msg.
> If, in future, we plan to use any other services from firmware, which is exposed in the extended TLVs, we will need to add a new variable (and not service_map_ext) to set the service.

I'm not sure I totally understood your response, but look at it from
the perspective of the function ath10k_wmi_event_service_available().

That function calls:

  ret = ath10k_wmi_pull_svc_avail(ar, skb, &arg);

...if it gets back a non-zero error code, it assumes that the
"arg.service_map_ext" and "arg.service_map_ext_len" values are now
valid and it can use them.

Before your patch, ath10k_wmi_pull_svc_avail() was returning an error
code.  That let ath10k_wmi_event_service_available() know that it
shouldn't look at "arg.service_map_ext" and "arg.service_map_ext_len".
After your patch, you're not returning an error code but those fields
aren't being filled in.

Said another way, if you remove the initialization of "arg" in
ath10k_wmi_event_service_available() then everything is broken.  While
things work because you _do_ have an initialization of "arg" in
ath10k_wmi_event_service_available(), it feels fragile to me to rely
on that.


-Doug