Re: [Intel-gfx] [PATCH 04/47] drm/i915/guc: Add non blocking CTB send function

[Matthew Brost <matthew.brost@intel.com>]

On Thu, Jun 24, 2021 at 07:02:18PM +0200, Michal Wajdeczko wrote:
> 
> 
> On 24.06.2021 17:49, Matthew Brost wrote:
> > On Thu, Jun 24, 2021 at 04:48:32PM +0200, Michal Wajdeczko wrote:
> >>
> >>
> >> On 24.06.2021 09:04, Matthew Brost wrote:
> >>> Add non blocking CTB send function, intel_guc_send_nb. GuC submission
> >>> will send CTBs in the critical path and does not need to wait for these
> >>> CTBs to complete before moving on, hence the need for this new function.
> >>>
> >>> The non-blocking CTB now must have a flow control mechanism to ensure
> >>> the buffer isn't overrun. A lazy spin wait is used as we believe the
> >>> flow control condition should be rare with a properly sized buffer.
> >>>
> >>> The function, intel_guc_send_nb, is exported in this patch but unused.
> >>> Several patches later in the series make use of this function.
> >>>
> >>> Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
> >>> Signed-off-by: Matthew Brost <matthew.brost@intel.com>
> >>> ---
> >>>  drivers/gpu/drm/i915/gt/uc/intel_guc.h    | 12 +++-
> >>>  drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c | 77 +++++++++++++++++++++--
> >>>  drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h |  3 +-
> >>>  3 files changed, 82 insertions(+), 10 deletions(-)
> >>>
> >>> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc.h b/drivers/gpu/drm/i915/gt/uc/intel_guc.h
> >>> index 4abc59f6f3cd..24b1df6ad4ae 100644
> >>> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc.h
> >>> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc.h
> >>> @@ -74,7 +74,15 @@ static inline struct intel_guc *log_to_guc(struct intel_guc_log *log)
> >>>  static
> >>>  inline int intel_guc_send(struct intel_guc *guc, const u32 *action, u32 len)
> >>>  {
> >>> -	return intel_guc_ct_send(&guc->ct, action, len, NULL, 0);
> >>> +	return intel_guc_ct_send(&guc->ct, action, len, NULL, 0, 0);
> >>> +}
> >>> +
> >>> +#define INTEL_GUC_SEND_NB		BIT(31)
> >>
> >> hmm, this flag really belongs to intel_guc_ct_send() so it should be
> >> defined as CTB flag near that function declaration
> >>
> > 
> > I can move this up a few lines.
> > 
> >>> +static
> >>> +inline int intel_guc_send_nb(struct intel_guc *guc, const u32 *action, u32 len)
> >>> +{
> >>> +	return intel_guc_ct_send(&guc->ct, action, len, NULL, 0,
> >>> +				 INTEL_GUC_SEND_NB);
> >>>  }
> >>>  
> >>>  static inline int
> >>> @@ -82,7 +90,7 @@ intel_guc_send_and_receive(struct intel_guc *guc, const u32 *action, u32 len,
> >>>  			   u32 *response_buf, u32 response_buf_size)
> >>>  {
> >>>  	return intel_guc_ct_send(&guc->ct, action, len,
> >>> -				 response_buf, response_buf_size);
> >>> +				 response_buf, response_buf_size, 0);
> >>>  }
> >>>  
> >>>  static inline void intel_guc_to_host_event_handler(struct intel_guc *guc)
> >>> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> >>> index a17215920e58..c9a65d05911f 100644
> >>> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> >>> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> >>> @@ -3,6 +3,11 @@
> >>>   * Copyright © 2016-2019 Intel Corporation
> >>>   */
> >>>  
> >>> +#include <linux/circ_buf.h>
> >>> +#include <linux/ktime.h>
> >>> +#include <linux/time64.h>
> >>> +#include <linux/timekeeping.h>
> >>> +
> >>>  #include "i915_drv.h"
> >>>  #include "intel_guc_ct.h"
> >>>  #include "gt/intel_gt.h"
> >>> @@ -373,7 +378,7 @@ static void write_barrier(struct intel_guc_ct *ct)
> >>>  static int ct_write(struct intel_guc_ct *ct,
> >>>  		    const u32 *action,
> >>>  		    u32 len /* in dwords */,
> >>> -		    u32 fence)
> >>> +		    u32 fence, u32 flags)
> >>>  {
> >>>  	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
> >>>  	struct guc_ct_buffer_desc *desc = ctb->desc;
> >>> @@ -421,9 +426,13 @@ static int ct_write(struct intel_guc_ct *ct,
> >>>  		 FIELD_PREP(GUC_CTB_MSG_0_NUM_DWORDS, len) |
> >>>  		 FIELD_PREP(GUC_CTB_MSG_0_FENCE, fence);
> >>>  
> >>> -	hxg = FIELD_PREP(GUC_HXG_MSG_0_TYPE, GUC_HXG_TYPE_REQUEST) |
> >>> -	      FIELD_PREP(GUC_HXG_REQUEST_MSG_0_ACTION |
> >>> -			 GUC_HXG_REQUEST_MSG_0_DATA0, action[0]);
> >>> +	hxg = (flags & INTEL_GUC_SEND_NB) ?
> >>> +		(FIELD_PREP(GUC_HXG_MSG_0_TYPE, GUC_HXG_TYPE_EVENT) |
> >>> +		 FIELD_PREP(GUC_HXG_EVENT_MSG_0_ACTION |
> >>> +			    GUC_HXG_EVENT_MSG_0_DATA0, action[0])) :
> >>> +		(FIELD_PREP(GUC_HXG_MSG_0_TYPE, GUC_HXG_TYPE_REQUEST) |
> >>> +		 FIELD_PREP(GUC_HXG_REQUEST_MSG_0_ACTION |
> >>> +			    GUC_HXG_REQUEST_MSG_0_DATA0, action[0]));
> >>
> >> or as we already switched to accept and return whole HXG messages in
> >> guc_send_mmio() maybe we should do the same for CTB variant too and
> >> instead of using extra flag just let caller to prepare proper HXG header
> >> with HXG_EVENT type and then in CTB code just look at this type to make
> >> decision which code path to use
> >>
> > 
> > Not sure I follow. Anyways could this be done in a follow up by you if
> > want this change.
> >  
> >> note that existing callers should not be impacted, as full HXG header
> >> for the REQUEST message looks exactly the same as "action" code alone.
> >>
> >>>  
> >>>  	CT_DEBUG(ct, "writing (tail %u) %*ph %*ph %*ph\n",
> >>>  		 tail, 4, &header, 4, &hxg, 4 * (len - 1), &action[1]);
> >>> @@ -498,6 +507,46 @@ static int wait_for_ct_request_update(struct ct_request *req, u32 *status)
> >>>  	return err;
> >>>  }
> >>>  
> >>> +static inline bool h2g_has_room(struct intel_guc_ct_buffer *ctb, u32 len_dw)
> >>> +{
> >>> +	struct guc_ct_buffer_desc *desc = ctb->desc;
> >>> +	u32 head = READ_ONCE(desc->head);
> >>> +	u32 space;
> >>> +
> >>> +	space = CIRC_SPACE(desc->tail, head, ctb->size);
> >>> +
> >>> +	return space >= len_dw;
> >>
> >> here you are returning true(1) as has room
> >>
> > 
> > See below.
> >  
> >>> +}
> >>> +
> >>> +static int ct_send_nb(struct intel_guc_ct *ct,
> >>> +		      const u32 *action,
> >>> +		      u32 len,
> >>> +		      u32 flags)
> >>> +{
> >>> +	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
> >>> +	unsigned long spin_flags;
> >>> +	u32 fence;
> >>> +	int ret;
> >>> +
> >>> +	spin_lock_irqsave(&ctb->lock, spin_flags);
> >>> +
> >>> +	ret = h2g_has_room(ctb, len + 1);
> >>
> >> but here you treat "1" it as en error
> >>
> > 
> > Yes, this patch is broken but fixed in a follow up one. Regardless I'll
> > fix this patch in place.
> > 
> >> and this "1" is GUC_HXG_MSG_MIN_LEN, right ?
> >>
> > 
> > Not exactly. This is following how ct_send() uses the action + len
> > field. Action[0] field goes in the HXG header and extra + 1 is for the
> > CT header.
> 
> well, "len" already counts "action" so by treating input as full HXG
> message (including HXG header) will make it cleaner
> 

Yes, I know. See above. To me GUC_HXG_MSG_MIN_LEN makes zero sense and
it is worse than adding + 1. This + 1 accounts for the CT header not the
HXG header. If any we add a new define, GUC_CT_HDR_LEN, and add that.

Matt

> we can try do it later but by doing it right now we would avoid
> introducing this send_nb() function and deprecating them long term again
> 
> > 
> >>> +	if (unlikely(ret))
> >>> +		goto out;
> >>> +
> >>> +	fence = ct_get_next_fence(ct);
> >>> +	ret = ct_write(ct, action, len, fence, flags);
> >>> +	if (unlikely(ret))
> >>> +		goto out;
> >>> +
> >>> +	intel_guc_notify(ct_to_guc(ct));
> >>> +
> >>> +out:
> >>> +	spin_unlock_irqrestore(&ctb->lock, spin_flags);
> >>> +
> >>> +	return ret;
> >>> +}
> >>> +
> >>>  static int ct_send(struct intel_guc_ct *ct,
> >>>  		   const u32 *action,
> >>>  		   u32 len,
> >>> @@ -505,6 +554,7 @@ static int ct_send(struct intel_guc_ct *ct,
> >>>  		   u32 response_buf_size,
> >>>  		   u32 *status)
> >>>  {
> >>> +	struct intel_guc_ct_buffer *ctb = &ct->ctbs.send;
> >>>  	struct ct_request request;
> >>>  	unsigned long flags;
> >>>  	u32 fence;
> >>> @@ -514,8 +564,20 @@ static int ct_send(struct intel_guc_ct *ct,
> >>>  	GEM_BUG_ON(!len);
> >>>  	GEM_BUG_ON(len & ~GUC_CT_MSG_LEN_MASK);
> >>>  	GEM_BUG_ON(!response_buf && response_buf_size);
> >>> +	might_sleep();
> >>>  
> >>> +	/*
> >>> +	 * We use a lazy spin wait loop here as we believe that if the CT
> >>> +	 * buffers are sized correctly the flow control condition should be
> >>> +	 * rare.
> >>
> >> shouldn't we at least try to log such cases with RATE_LIMITED to find
> >> out how "rare" it is, or if really unlikely just return -EBUSY as in
> >> case of non-blocking send ?
> >>
> > 
> > Definitely not return -EBUSY as this a blocking call. Perhaps we can log
> 
> blocking calls still can fail for various reasons, full CTB is one of
> them, and if we return error (now broken) for non-blocking variant then
> we should do the same for blocking variant as well and let the caller
> decide about next steps
> 
> > this, but IGTs likely can hit rather easily. It really is only
> > interesting if real workloads hit this. Regardless that can be a follow
> > up.
> 
> if we hide retry in a silent loop then we will not find it out if we hit
> this condition (IGT or real WL) or not
> 
> > 
> > Matt
> >  
> >>> +	 */
> >>> +retry:
> >>>  	spin_lock_irqsave(&ct->ctbs.send.lock, flags);
> >>> +	if (unlikely(!h2g_has_room(ctb, len + 1))) {
> >>> +		spin_unlock_irqrestore(&ct->ctbs.send.lock, flags);
> >>> +		cond_resched();
> >>> +		goto retry;
> >>> +	}
> >>>  
> >>>  	fence = ct_get_next_fence(ct);
> >>>  	request.fence = fence;
> >>> @@ -527,7 +589,7 @@ static int ct_send(struct intel_guc_ct *ct,
> >>>  	list_add_tail(&request.link, &ct->requests.pending);
> >>>  	spin_unlock(&ct->requests.lock);
> >>>  
> >>> -	err = ct_write(ct, action, len, fence);
> >>> +	err = ct_write(ct, action, len, fence, 0);
> >>>  
> >>>  	spin_unlock_irqrestore(&ct->ctbs.send.lock, flags);
> >>>  
> >>> @@ -569,7 +631,7 @@ static int ct_send(struct intel_guc_ct *ct,
> >>>   * Command Transport (CT) buffer based GuC send function.
> >>>   */
> >>>  int intel_guc_ct_send(struct intel_guc_ct *ct, const u32 *action, u32 len,
> >>> -		      u32 *response_buf, u32 response_buf_size)
> >>> +		      u32 *response_buf, u32 response_buf_size, u32 flags)
> >>>  {
> >>>  	u32 status = ~0; /* undefined */
> >>>  	int ret;
> >>> @@ -579,6 +641,9 @@ int intel_guc_ct_send(struct intel_guc_ct *ct, const u32 *action, u32 len,
> >>>  		return -ENODEV;
> >>>  	}
> >>>  
> >>> +	if (flags & INTEL_GUC_SEND_NB)
> >>> +		return ct_send_nb(ct, action, len, flags);
> >>> +
> >>>  	ret = ct_send(ct, action, len, response_buf, response_buf_size, &status);
> >>>  	if (unlikely(ret < 0)) {
> >>>  		CT_ERROR(ct, "Sending action %#x failed (err=%d status=%#X)\n",
> >>> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> >>> index 1ae2dde6db93..eb69263324ba 100644
> >>> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> >>> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.h
> >>> @@ -42,7 +42,6 @@ struct intel_guc_ct_buffer {
> >>>  	bool broken;
> >>>  };
> >>>  
> >>> -
> >>>  /** Top-level structure for Command Transport related data
> >>>   *
> >>>   * Includes a pair of CT buffers for bi-directional communication and tracking
> >>> @@ -88,7 +87,7 @@ static inline bool intel_guc_ct_enabled(struct intel_guc_ct *ct)
> >>>  }
> >>>  
> >>>  int intel_guc_ct_send(struct intel_guc_ct *ct, const u32 *action, u32 len,
> >>> -		      u32 *response_buf, u32 response_buf_size);
> >>> +		      u32 *response_buf, u32 response_buf_size, u32 flags);
> >>>  void intel_guc_ct_event_handler(struct intel_guc_ct *ct);
> >>>  
> >>>  #endif /* _INTEL_GUC_CT_H_ */
> >>>