[Public] Hi Jani, Thanks for your time! Comments inline. > -----Original Message----- > From: Jani Nikula <jani.nikula@xxxxxxxxx> > Sent: Tuesday, May 16, 2023 6:34 PM > To: Lin, Wayne <Wayne.Lin@xxxxxxx>; dri-devel@xxxxxxxxxxxxxxxxxxxxx; > amd-gfx@xxxxxxxxxxxxxxxxxxxxx > Cc: lyude@xxxxxxxxxx; ville.syrjala@xxxxxxxxxxxxxxx; imre.deak@xxxxxxxxx; > Wentland, Harry <Harry.Wentland@xxxxxxx>; Zuo, Jerry > <Jerry.Zuo@xxxxxxx>; Lin, Wayne <Wayne.Lin@xxxxxxx>; > stable@xxxxxxxxxxxxxxx > Subject: Re: [PATCH v2] drm/dp_mst: Clear MSG_RDY flag before sending new > message > > On Thu, 27 Apr 2023, Wayne Lin <Wayne.Lin@xxxxxxx> wrote: > > [Why] > > The sequence for collecting down_reply from source perspective should > > be: > > > > Request_n->repeat (get partial reply of Request_n->clear message ready > > flag to ack DPRX that the message is received) till all partial > > replies for Request_n are received->new Request_n+1. > > > > Now there is chance that drm_dp_mst_hpd_irq() will fire new down > > request in the tx queue when the down reply is incomplete. Source is > > restricted to generate interveleaved message transactions so we should > > avoid it. > > > > Also, while assembling partial reply packets, reading out DPCD > > DOWN_REP Sideband MSG buffer + clearing DOWN_REP_MSG_RDY flag > should > > be wrapped up as a complete operation for reading out a reply packet. > > Kicking off a new request before clearing DOWN_REP_MSG_RDY flag might > > be risky. e.g. If the reply of the new request has overwritten the > > DPRX DOWN_REP Sideband MSG buffer before source writing one to clear > > DOWN_REP_MSG_RDY flag, source then unintentionally flushes the reply > > for the new request. Should handle the up request in the same way. > > > > [How] > > Separete drm_dp_mst_hpd_irq() into 2 steps. After acking the MST IRQ > > event, driver calls drm_dp_mst_hpd_irq_step2() and might trigger > > drm_dp_mst_kick_tx() only when there is no on going message transaction. > > > > Changes since v1: > > * Reworked on review comments received > > -> Adjust the fix to let driver explicitly kick off new down request > > when mst irq event is handled and acked > > -> Adjust the commit message > > > > Signed-off-by: Wayne Lin <Wayne.Lin@xxxxxxx> > > Cc: stable@xxxxxxxxxxxxxxx > > --- > > .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 8 ++--- > > drivers/gpu/drm/display/drm_dp_mst_topology.c | 35 > ++++++++++++++++--- > > drivers/gpu/drm/i915/display/intel_dp.c | 5 ++- > > drivers/gpu/drm/nouveau/dispnv50/disp.c | 5 ++- > > include/drm/display/drm_dp_mst_helper.h | 4 +-- > > 5 files changed, 45 insertions(+), 12 deletions(-) > > > > diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c > > b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c > > index 1ad67c2a697e..48bdcb2ee9b1 100644 > > --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c > > +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c > > @@ -3259,10 +3259,9 @@ static void > dm_handle_mst_sideband_msg(struct amdgpu_dm_connector *aconnector) > > DRM_DEBUG_DRIVER("ESI %02x %02x %02x\n", esi[0], > esi[1], esi[2]); > > /* handle HPD short pulse irq */ > > if (aconnector->mst_mgr.mst_state) > > - drm_dp_mst_hpd_irq( > > - &aconnector->mst_mgr, > > - esi, > > - &new_irq_handled); > > + drm_dp_mst_hpd_irq_step1(&aconnector->mst_mgr, > > + esi, > > + &new_irq_handled); > > > > if (new_irq_handled) { > > /* ACK at DPCD to notify down stream */ @@ - > 3281,6 +3280,7 @@ > > static void dm_handle_mst_sideband_msg(struct amdgpu_dm_connector > *aconnector) > > break; > > } > > > > + drm_dp_mst_hpd_irq_step2(&aconnector- > >mst_mgr); > > /* check if there is new irq to be handled */ > > dret = drm_dp_dpcd_read( > > &aconnector->dm_dp_aux.aux, > > diff --git a/drivers/gpu/drm/display/drm_dp_mst_topology.c > > b/drivers/gpu/drm/display/drm_dp_mst_topology.c > > index 70df29fe92db..2e0a38a6509c 100644 > > --- a/drivers/gpu/drm/display/drm_dp_mst_topology.c > > +++ b/drivers/gpu/drm/display/drm_dp_mst_topology.c > > @@ -4045,7 +4045,7 @@ static int drm_dp_mst_handle_up_req(struct > > drm_dp_mst_topology_mgr *mgr) } > > > > /** > > - * drm_dp_mst_hpd_irq() - MST hotplug IRQ notify > > + * drm_dp_mst_hpd_irq_step1() - MST hotplug IRQ notify > > * @mgr: manager to notify irq for. > > * @esi: 4 bytes from SINK_COUNT_ESI > > * @handled: whether the hpd interrupt was consumed or not @@ -4055,7 > > +4055,7 @@ static int drm_dp_mst_handle_up_req(struct > drm_dp_mst_topology_mgr *mgr) > > * topology manager will process the sideband messages received as a result > > * of this. > > */ > > -int drm_dp_mst_hpd_irq(struct drm_dp_mst_topology_mgr *mgr, u8 *esi, > > bool *handled) > > +int drm_dp_mst_hpd_irq_step1(struct drm_dp_mst_topology_mgr *mgr, > u8 > > +*esi, bool *handled) > > If you're changing the signature of the function, I'd make esi "const u8 *esi", > and add a separate "u8 *ack" that you have to provide, where this function > would |= the flags to ack. It would be useful at least in i915. Will adjust. Thanks. > > As to naming, _step1 and _step2 are pretty vague. Was trying to align the naming method we used for payload allocation/de-allocation. Anyway, I'll adjust the naming here. > > > { > > int ret = 0; > > int sc; > > @@ -4077,11 +4077,38 @@ int drm_dp_mst_hpd_irq(struct > drm_dp_mst_topology_mgr *mgr, u8 *esi, bool *handl > > *handled = true; > > } > > > > - drm_dp_mst_kick_tx(mgr); > > return ret; > > } > > -EXPORT_SYMBOL(drm_dp_mst_hpd_irq); > > +EXPORT_SYMBOL(drm_dp_mst_hpd_irq_step1); > > + > > +/** > > + * drm_dp_mst_hpd_irq_step2() - MST hotplug IRQ 2nd part handling > > + * @mgr: manager to notify irq for. > > + * > > + * This should be called from the driver when mst irq event is > > +handled > > + * and acked. Note that new down request should only be sent when > > + * previous message transaction is done. Source is not supposed to > > +generate > > + * interleaved message transactions. > > + */ > > +void drm_dp_mst_hpd_irq_step2(struct drm_dp_mst_topology_mgr > *mgr) > > _done, _finish, _complete? Will use "complete". Thanks. > > > +{ > > + struct drm_dp_sideband_msg_tx *txmsg; > > + bool skip = false; > > > > + mutex_lock(&mgr->qlock); > > + txmsg = list_first_entry_or_null(&mgr->tx_msg_downq, > > + struct drm_dp_sideband_msg_tx, > next); > > + /* If last transaction is not completed yet*/ > > + if (!txmsg || > > + txmsg->state == DRM_DP_SIDEBAND_TX_START_SEND || > > + txmsg->state == DRM_DP_SIDEBAND_TX_SENT) > > + skip = true; > > + mutex_unlock(&mgr->qlock); > > + > > + if (!skip) > > Please avoid negatives like this. You could have bool kick = true instead. Thanks. Will modify it. > > > + drm_dp_mst_kick_tx(mgr); > > +} > > +EXPORT_SYMBOL(drm_dp_mst_hpd_irq_step2); > > /** > > * drm_dp_mst_detect_port() - get connection status for an MST port > > * @connector: DRM connector for this port diff --git > > a/drivers/gpu/drm/i915/display/intel_dp.c > > b/drivers/gpu/drm/i915/display/intel_dp.c > > index 75070eb07d4b..9a9a5aec9534 100644 > > --- a/drivers/gpu/drm/i915/display/intel_dp.c > > +++ b/drivers/gpu/drm/i915/display/intel_dp.c > > @@ -3803,7 +3803,7 @@ intel_dp_mst_hpd_irq(struct intel_dp *intel_dp, > > u8 *esi, u8 *ack) { > > bool handled = false; > > > > - drm_dp_mst_hpd_irq(&intel_dp->mst_mgr, esi, &handled); > > + drm_dp_mst_hpd_irq_step1(&intel_dp->mst_mgr, esi, &handled); > > if (handled) > > ack[1] |= esi[1] & (DP_DOWN_REP_MSG_RDY | > DP_UP_REQ_MSG_RDY); > > > > @@ -3880,6 +3880,9 @@ intel_dp_check_mst_status(struct intel_dp > > *intel_dp) > > > > if (!intel_dp_ack_sink_irq_esi(intel_dp, ack)) > > drm_dbg_kms(&i915->drm, "Failed to ack ESI\n"); > > + > > + if (ack[1] & (DP_DOWN_REP_MSG_RDY | > DP_UP_REQ_MSG_RDY)) > > + drm_dp_mst_hpd_irq_step2(&intel_dp->mst_mgr); > > I'm getting confused about the division of responsibilities between the two > functions to be called, and the caller. Why does i915 do things differently from > nouveau and amd wrt this? The main idea is trying to ack the irq before sending a new request. We used to send a new request before acking the irq event which will cause message interleaving and that's not recommended by DP spec. amd and nouveau only handle mst up/down message irq events while calling drm_dp_mst_hpd_irq() but i915 also tries to handle content protection irq while handling mst events. That's why it's a bit different between amd/nouveau and i915. > > > } > > > > return link_ok; > > diff --git a/drivers/gpu/drm/nouveau/dispnv50/disp.c > > b/drivers/gpu/drm/nouveau/dispnv50/disp.c > > index ed9d374147b8..00c36fcc8afd 100644 > > --- a/drivers/gpu/drm/nouveau/dispnv50/disp.c > > +++ b/drivers/gpu/drm/nouveau/dispnv50/disp.c > > @@ -1332,12 +1332,15 @@ nv50_mstm_service(struct nouveau_drm > *drm, > > break; > > } > > > > - drm_dp_mst_hpd_irq(&mstm->mgr, esi, &handled); > > + drm_dp_mst_hpd_irq_step1(&mstm->mgr, esi, &handled); > > if (!handled) > > break; > > > > rc = drm_dp_dpcd_write(aux, DP_SINK_COUNT_ESI + 1, > &esi[1], > > 3); > > + > > + drm_dp_mst_hpd_irq_step2(&mstm->mgr); > > + > > Don't you think the return value should be checked first? Sorry I missed it. Will adjust. Thanks! > > > if (rc != 3) { > > ret = false; > > break; > > diff --git a/include/drm/display/drm_dp_mst_helper.h > > b/include/drm/display/drm_dp_mst_helper.h > > index 32c764fb9cb5..6c08ba765d5a 100644 > > --- a/include/drm/display/drm_dp_mst_helper.h > > +++ b/include/drm/display/drm_dp_mst_helper.h > > @@ -815,8 +815,8 @@ void drm_dp_mst_topology_mgr_destroy(struct > > drm_dp_mst_topology_mgr *mgr); bool drm_dp_read_mst_cap(struct > > drm_dp_aux *aux, const u8 dpcd[DP_RECEIVER_CAP_SIZE]); int > > drm_dp_mst_topology_mgr_set_mst(struct drm_dp_mst_topology_mgr > *mgr, > > bool mst_state); > > > > -int drm_dp_mst_hpd_irq(struct drm_dp_mst_topology_mgr *mgr, u8 *esi, > > bool *handled); > > - > > +int drm_dp_mst_hpd_irq_step1(struct drm_dp_mst_topology_mgr *mgr, > u8 > > +*esi, bool *handled); void drm_dp_mst_hpd_irq_step2(struct > > +drm_dp_mst_topology_mgr *mgr); > > > > int > > drm_dp_mst_detect_port(struct drm_connector *connector, > > -- > Jani Nikula, Intel Open Source Graphics Center -- Regards, Wayne Lin