Date:   Fri, 1 Oct 2021 22:58:29 +0000
From:   <Kelvin.Cao@...rochip.com>
To:     <helgaas@...nel.org>
CC:     <kurt.schwemmer@...rosemi.com>, <bhelgaas@...gle.com>,
        <kelvincao@...look.com>, <logang@...tatee.com>,
        <linux-kernel@...r.kernel.org>, <linux-pci@...r.kernel.org>
Subject: Re: [PATCH 1/5] PCI/switchtec: Error out MRPC execution when no GAS
 access

On Fri, 2021-10-01 at 15:18 -0500, Bjorn Helgaas wrote:
> On Fri, Sep 24, 2021 at 11:08:38AM +0000, kelvin.cao@...rochip.com
> wrote:
> > From: Kelvin Cao <kelvin.cao@...rochip.com>
> > 
> > After a firmware hard reset, MRPC command executions, which are
> > based
> > on the PCI BAR (which Microchip refers to as GAS) read/write, will
> > hang
> > indefinitely. This is because after a reset, the host will fail all
> > GAS
> > reads (get all 1s), in which case the driver won't get a valid MRPC
> > status.
> 
> Trying to write a merge commit log for this, but having a hard time
> summarizing it.  It sounds like it covers both Switchtec-specific
> (firmware and MRPC commands) and generic PCIe behavior (MMIO read
> failures).
> 
> This has something to do with a firmware hard reset.  What is that?
> Is that like a firmware reboot?  A device reset, e.g., FLR or
> secondary bus reset, that causes a firmware reboot?  A device reset
> initiated by firmware?
> 
> Anyway, apparently when that happens, MMIO reads to the switch fail
> (timeout or error completion on PCIe) for a while.  If a device reset
> is involved, that much is standard PCIe behavior.  And the driver
> sees
> ~0 data from those failed reads.  That's not part of the PCIe spec,
> but is typical root complex behavior.
> 
> But you said the MRPC commands hang indefinitely.  Presumably MMIO
> reads would start succeeding eventually when the device becomes
> ready,
> so I don't know how that translates to "indefinitely."
> 
> Weird to refer to a PCI BAR as "GAS".  Maybe expanding the acronym
> would help it make sense.
> 
> What does "host" refer to?  I guess it's the switch (the
> switchtec_dev), since you say it fails MMIO reads?
> 
> Naming comment below.
> 
> > Add a read check to GAS access when a MRPC command execution
> > doesn't
> > response timely, error out if the check fails.
> > 
> > Signed-off-by: Kelvin Cao <kelvin.cao@...rochip.com>
> > ---
> >  drivers/pci/switch/switchtec.c | 59
> > ++++++++++++++++++++++++++++++----
> >  1 file changed, 52 insertions(+), 7 deletions(-)
> > 
> > diff --git a/drivers/pci/switch/switchtec.c
> > b/drivers/pci/switch/switchtec.c
> > index 0b301f8be9ed..092653487021 100644
> > --- a/drivers/pci/switch/switchtec.c
> > +++ b/drivers/pci/switch/switchtec.c
> > @@ -45,6 +45,7 @@ enum mrpc_state {
> >       MRPC_QUEUED,
> >       MRPC_RUNNING,
> >       MRPC_DONE,
> > +     MRPC_IO_ERROR,
> >  };
> > 
> >  struct switchtec_user {
> > @@ -66,6 +67,13 @@ struct switchtec_user {
> >       int event_cnt;
> >  };
> > 
> > +static int check_access(struct switchtec_dev *stdev)
> > +{
> > +     u32 device = ioread32(&stdev->mmio_sys_info->device_id);
> > +
> > +     return stdev->pdev->device == device;
> > +}
> 
> Didn't notice this before, but the "check_access()" name is not very
> helpful because it doesn't give a clue about what the return value
> means.  Does 0 mean no error?  Does 1 mean no error?  From reading
> the
> implementation, I can see that 0 is actually the error case, but I
> can't tell from the name.

Yes, will improve the naming, like change it to "has_gas_access()" in
v2 if a v2 patchset is preferred.
> 
> >  static struct switchtec_user *stuser_create(struct switchtec_dev
> > *stdev)
> >  {
> >       struct switchtec_user *stuser;
> > @@ -113,6 +121,7 @@ static void stuser_set_state(struct
> > switchtec_user *stuser,
> >               [MRPC_QUEUED] = "QUEUED",
> >               [MRPC_RUNNING] = "RUNNING",
> >               [MRPC_DONE] = "DONE",
> > +             [MRPC_IO_ERROR] = "IO_ERROR",
> >       };
> > 
> >       stuser->state = state;
> > @@ -184,6 +193,21 @@ static int mrpc_queue_cmd(struct
> > switchtec_user *stuser)
> >       return 0;
> >  }
> > 
> > +static void mrpc_cleanup_cmd(struct switchtec_dev *stdev)
> > +{
> > +     /* requires the mrpc_mutex to already be held when called */
> > +     struct switchtec_user *stuser = list_entry(stdev-
> > >mrpc_queue.next,
> > +                                                struct
> > switchtec_user, list);
> > +
> > +     stuser->cmd_done = true;
> > +     wake_up_interruptible(&stuser->cmd_comp);
> > +     list_del_init(&stuser->list);
> > +     stuser_put(stuser);
> > +     stdev->mrpc_busy = 0;
> > +
> > +     mrpc_cmd_submit(stdev);
> > +}
> > +
> >  static void mrpc_complete_cmd(struct switchtec_dev *stdev)
> >  {
> >       /* requires the mrpc_mutex to already be held when called */
> > @@ -223,13 +247,7 @@ static void mrpc_complete_cmd(struct
> > switchtec_dev *stdev)
> >               memcpy_fromio(stuser->data, &stdev->mmio_mrpc-
> > >output_data,
> >                             stuser->read_len);
> >  out:
> > -     stuser->cmd_done = true;
> > -     wake_up_interruptible(&stuser->cmd_comp);
> > -     list_del_init(&stuser->list);
> > -     stuser_put(stuser);
> > -     stdev->mrpc_busy = 0;
> > -
> > -     mrpc_cmd_submit(stdev);
> > +     mrpc_cleanup_cmd(stdev);
> >  }
> > 
> >  static void mrpc_event_work(struct work_struct *work)
> > @@ -246,6 +264,23 @@ static void mrpc_event_work(struct work_struct
> > *work)
> >       mutex_unlock(&stdev->mrpc_mutex);
> >  }
> > 
> > +static void mrpc_error_complete_cmd(struct switchtec_dev *stdev)
> > +{
> > +     /* requires the mrpc_mutex to already be held when called */
> > +
> > +     struct switchtec_user *stuser;
> > +
> > +     if (list_empty(&stdev->mrpc_queue))
> > +             return;
> > +
> > +     stuser = list_entry(stdev->mrpc_queue.next,
> > +                         struct switchtec_user, list);
> > +
> > +     stuser_set_state(stuser, MRPC_IO_ERROR);
> > +
> > +     mrpc_cleanup_cmd(stdev);
> > +}
> > +
> >  static void mrpc_timeout_work(struct work_struct *work)
> >  {
> >       struct switchtec_dev *stdev;
> > @@ -257,6 +292,11 @@ static void mrpc_timeout_work(struct
> > work_struct *work)
> > 
> >       mutex_lock(&stdev->mrpc_mutex);
> > 
> > +     if (!check_access(stdev)) {
> > +             mrpc_error_complete_cmd(stdev);
> > +             goto out;
> > +     }
> > +
> >       if (stdev->dma_mrpc)
> >               status = stdev->dma_mrpc->status;
> >       else
> > @@ -544,6 +584,11 @@ static ssize_t switchtec_dev_read(struct file
> > *filp, char __user *data,
> >       if (rc)
> >               return rc;
> > 
> > +     if (stuser->state == MRPC_IO_ERROR) {
> > +             mutex_unlock(&stdev->mrpc_mutex);
> > +             return -EIO;
> > +     }
> > +
> >       if (stuser->state != MRPC_DONE) {
> >               mutex_unlock(&stdev->mrpc_mutex);
> >               return -EBADE;
> > --
> > 2.25.1
> > 

Powered by blists - more mailing lists