<html> <head> <meta http-equiv="Content-Type" content="text/html; charset=us-ascii"> <style type="text/css" style="display:none;"></style> </head> <body dir="ltr"> <div id="divtagdefaultwrapper" style="font-size:12pt;color:#000000;font-family:Calibri,Helvetica,sans-serif;" dir="ltr"> <p style="margin-top:0;margin-bottom:0">okay</p> </div> <hr style="display:inline-block;width:98%" tabindex="-1"> <div id="divRplyFwdMsg" dir="ltr"><font face="Calibri, sans-serif" style="font-size:11pt" color="#000000"><b>From:</b> Chris Wilson <chris@chris-wilson.co.uk><br> <b>Sent:</b> Tuesday, March 6, 2018 4:24:21 PM<br> <b>To:</b> Liu, Monk; dri-devel@freedesktop.org<br> <b>Cc:</b> Liu, Monk<br> <b>Subject:</b> Re: [PATCH] dma-buf/reservation: should keep later one in add fence(v2)</font> <div> </div> </div> <div class="BodyFragment"><font size="2"><span style="font-size:11pt;"> <div class="PlainText">Quoting Monk Liu (2018-03-06 03:53:10)<br> > v2:<br> > still check context first to avoid warning from dma_fence_is_later<br> > apply this fix in add_shared_replace as well<br> > <br> > Change-Id: If6a979ba9fd6c923b82212f35f07a9ff31c86767<br> > Signed-off-by: Monk Liu <Monk.Liu@amd.com><br> > ---<br> > drivers/dma-buf/reservation.c | 6 ++++--<br> > 1 file changed, 4 insertions(+), 2 deletions(-)<br> > <br> > diff --git a/drivers/dma-buf/reservation.c b/drivers/dma-buf/reservation.c<br> > index 314eb10..c6e3c86 100644<br> > --- a/drivers/dma-buf/reservation.c<br> > +++ b/drivers/dma-buf/reservation.c<br> > @@ -118,7 +118,8 @@ reservation_object_add_shared_inplace(struct reservation_object *obj,<br> > old_fence = rcu_dereference_protected(fobj->shared[i],<br> > reservation_object_held(obj));<br> > <br> > - if (old_fence->context == fence->context) {<br> > + if (old_fence->context == fence->context &&<br> > + dma_fence_is_later(fence, old_fence)) {<br> <br> This should be true by construction. Adding an older fence on the same<br> context is a programming bug, imo. Between different callers the resv<br> should have been locked and the fenced operations serialised, from the<br> same caller, you shouldn't be handling more than one output fence?<br> -Chris<br> </div> </span></font></div> </body> </html>