Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755185AbcJMNez (ORCPT ); Thu, 13 Oct 2016 09:34:55 -0400 Received: from mga09.intel.com ([134.134.136.24]:3666 "EHLO mga09.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753231AbcJMNeo (ORCPT ); Thu, 13 Oct 2016 09:34:44 -0400 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.31,339,1473145200"; d="scan'208";a="1044191898" From: Felipe Balbi To: Baolin Wang Cc: Greg KH , Mark Brown , USB , LKML Subject: Re: [PATCH v2] usb: dwc3: gadget: Wait for end transfer complete before free irq In-Reply-To: References: <87r37kviic.fsf@linux.intel.com> <87inswvg80.fsf@linux.intel.com> <87a8e8vaif.fsf@linux.intel.com> <877f9cv7no.fsf@linux.intel.com> <87y41struu.fsf@linux.intel.com> Date: Thu, 13 Oct 2016 16:34:08 +0300 Message-ID: <87shs0tltb.fsf@linux.intel.com> MIME-Version: 1.0 Content-Type: text/plain Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 7333 Lines: 213 Hi, Baolin Wang writes: >> Baolin Wang writes: >>>>> Baolin Wang writes: >>>>>>>>> I'm thinking this is a bug in configfs interface of Gadget API, not >>>>>>>>> dwc3. The only reason for this to happen would be if we get to >>>>>>>>> ->udc_stop() with endpoints still enabled. >>>>>>>>> >>>>>>>>> Can you check if that's the case? i.e. can you check if any endpoints >>>>>>>>> are still enabled when we get here? >>>>>>>> >>>>>>>> No, it is not any endpoints are still enabled. Like I said in commit >>>>>>>> message, we will start end transfer command when disable the endpoint, >>>>>>>> if the end transfer command complete event comes after we have freed >>>>>>>> the gadget irq, it will trigger the interrupt line all the time to >>>>>>>> crash the system. >>>>>>> >>>>>>> I see what the problem is. Databook tells us we *must* set CMDIOC when >>>>>>> issuing EndTransfer command and we should always wait for Command >>>>>>> Complete IRQ. However, we took a shortcut and just delayed for 100us >>>>>>> after issuing End Transfer. >>>>>> >>>>>> Yes, but 100us delay is not enough in some scenarios, like changing >>>>>> function with configfs frequently, we will met problems. >>>>> >>>>> heh, 100us *is* enough. However we still get an IRQ because we requested >>>>> for it. If you want to fix this, then you need to find a way to get rid >>>>> of that 100us udelay() and add a proper wait_for_completion() to delay >>>>> execution until command complete IRQ fires up. >>>> >>>> I haven't tested this yet, but it shows the idea (I think we might still >>>> have a race with ep_queue if we're still waiting for End Transfer, but >>> >>> Yes, maybe we need check DWC3_EP_END_TRANSFER_PENDING flag when >>> queuing one request. >> >> Yeah, I'll add that check later. I still need to make sure we would >> still try to kick any transfers that might have been queued while >> waiting for End Transfer Command Complete IRQ. > > OK. But I still worried if there are other races in some places which There are no other places where this needs to be sorted out. > is not easy to find out by testing. No introducing race condition > would be one better solution, anyway I would like to test the patch > you send out firstly. Right, but your patch was working around another problem, rather then fixing it. Here's another version which should make sure everything remains working as it should. 8<------------------------------------------------------------------------ >From 1f922a902a21d5cee32082be18bbfbf1d46137e4 Mon Sep 17 00:00:00 2001 From: Felipe Balbi Date: Thu, 13 Oct 2016 14:09:47 +0300 Subject: [PATCH] usb: dwc3: gadget: wait for End Transfer to complete Instead of just delaying for 100us, we should actually wait for End Transfer Command Complete interrupt before moving on. Note that this should only be done if we're dealing with one of the core revisions that actually require the interrupt before moving on. Reported-by: Baolin Wang Signed-off-by: Felipe Balbi --- drivers/usb/dwc3/core.h | 10 +++++++++- drivers/usb/dwc3/gadget.c | 31 ++++++++++++++++++++++++++++--- 2 files changed, 37 insertions(+), 4 deletions(-) diff --git a/drivers/usb/dwc3/core.h b/drivers/usb/dwc3/core.h index e878366ead00..cf495d932252 100644 --- a/drivers/usb/dwc3/core.h +++ b/drivers/usb/dwc3/core.h @@ -26,6 +26,7 @@ #include #include #include +#include #include #include @@ -504,6 +505,7 @@ struct dwc3_event_buffer { * @endpoint: usb endpoint * @pending_list: list of pending requests for this endpoint * @started_list: list of started requests on this endpoint + * @wait_end_transfer: wait_queue_head_t for waiting on End Transfer complete * @lock: spinlock for endpoint request queue traversal * @regs: pointer to first endpoint register * @trb_pool: array of transaction buffers @@ -529,6 +531,8 @@ struct dwc3_ep { struct list_head pending_list; struct list_head started_list; + wait_queue_head_t wait_end_transfer; + spinlock_t lock; void __iomem *regs; @@ -545,7 +549,8 @@ struct dwc3_ep { #define DWC3_EP_BUSY (1 << 4) #define DWC3_EP_PENDING_REQUEST (1 << 5) #define DWC3_EP_MISSED_ISOC (1 << 6) - +#define DWC3_EP_END_TRANSFER_PENDING (1 << 7) +#define DWC3_EP_KICK_POST_END_TRANSFER (1 << 8) /* This last one is specific to EP0 */ #define DWC3_EP0_DIR_IN (1 << 31) @@ -1047,6 +1052,9 @@ struct dwc3_event_depevt { #define DEPEVT_TRANSFER_BUS_EXPIRY 2 u32 parameters:16; + +/* For Command Complete Events */ +#define DEPEVT_PARAMETER_CMD(n) (((n) & (0xf << 24)) >> 24) } __packed; /** diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c index 3ba05b12e49a..a3f81b5ba901 100644 --- a/drivers/usb/dwc3/gadget.c +++ b/drivers/usb/dwc3/gadget.c @@ -598,6 +598,8 @@ static int __dwc3_gadget_ep_enable(struct dwc3_ep *dep, reg |= DWC3_DALEPENA_EP(dep->number); dwc3_writel(dwc->regs, DWC3_DALEPENA, reg); + init_waitqueue_head(&dep->wait_end_transfer); + if (usb_endpoint_xfer_control(desc)) return 0; @@ -1151,6 +1153,11 @@ static int __dwc3_gadget_ep_queue(struct dwc3_ep *dep, struct dwc3_request *req) if (!dwc3_calc_trbs_left(dep)) return 0; + if (dep->flags & DWC3_EP_END_TRANSFER_PENDING) { + dep->flags &= DWC3_EP_KICK_POST_END_TRANSFER; + return 0; + } + ret = __dwc3_gadget_kick_transfer(dep, 0); if (ret && ret != -EBUSY) dwc3_trace(trace_dwc3_gadget, @@ -2156,6 +2163,7 @@ static void dwc3_endpoint_interrupt(struct dwc3 *dwc, { struct dwc3_ep *dep; u8 epnum = event->endpoint_number; + u8 cmd; dep = dwc->eps[epnum]; @@ -2200,8 +2208,13 @@ static void dwc3_endpoint_interrupt(struct dwc3 *dwc, return; } break; - case DWC3_DEPEVT_RXTXFIFOEVT: case DWC3_DEPEVT_EPCMDCMPLT: + cmd = DEPEVT_PARAMETER_CMD(event->parameters); + + if (cmd == DWC3_DEPCMD_ENDTRANSFER) + dep->flags &= ~DWC3_EP_END_TRANSFER_PENDING; + break; + case DWC3_DEPEVT_RXTXFIFOEVT: break; } } @@ -2246,6 +2259,7 @@ static void dwc3_reset_gadget(struct dwc3 *dwc) } static void dwc3_stop_active_transfer(struct dwc3 *dwc, u32 epnum, bool force) +__releases(&dwc->lock) __acquires(&dwc->lock) { struct dwc3_ep *dep; struct dwc3_gadget_ep_cmd_params params; @@ -2295,11 +2309,22 @@ static void dwc3_stop_active_transfer(struct dwc3 *dwc, u32 epnum, bool force) memset(¶ms, 0, sizeof(params)); ret = dwc3_send_gadget_ep_cmd(dep, cmd, ¶ms); WARN_ON_ONCE(ret); + + if (dwc3_is_usb31(dwc) || dwc->revision < DWC3_REVISION_310A) { + dep->flags |= DWC3_EP_END_TRANSFER_PENDING; + wait_event_lock_irq(dep->wait_end_transfer, + !(dep->flags & DWC3_EP_END_TRANSFER_PENDING), + dwc->lock); + } + dep->resource_index = 0; dep->flags &= ~DWC3_EP_BUSY; - if (dwc3_is_usb31(dwc) || dwc->revision < DWC3_REVISION_310A) - udelay(100); + if (dep->flags & DWC3_EP_KICK_POST_END_TRANSFER) { + dep->flags &= ~DWC3_EP_KICK_POST_END_TRANSFER; + ret = __dwc3_gadget_kick_transfer(dep, 0); + WARN_ON_ONCE(ret); + } } static void dwc3_stop_active_transfers(struct dwc3 *dwc) -- 2.10.0.440.g21f862b -- balbi