- 04 Mar, 2016 40 commits
-
-
Douglas Anderson authored
When setting up ISO and INT transfers dwc2 needs to specify whether the transfer is for an even or an odd frame (or microframe if the controller is running in high speed mode). The controller appears to use this as a simple way to figure out if a transfer should happen right away (in the current microframe) or should happen at the start of the next microframe. Said another way: - If you set "odd" and the current frame number is odd it appears that the controller will try to transfer right away. Same thing if you set "even" and the current frame number is even. - If the oddness you set and the oddness of the frame number are _different_, the transfer will be delayed until the frame number changes. As I understand it, the above technique allows you to plan ahead of time where possible by always working on the next frame. ...but it still allows you to properly respond immediately to things that happened in the previous frame. The old dwc2_hc_set_even_odd_frame() didn't really handle this concept. It always looked at the frame number and setup the transfer to happen in the next frame. In some cases that meant that certain transactions would be transferred in the wrong frame. We'll try our best to set the even / odd to do the transfer in the scheduled frame. If that fails then we'll do an ugly "schedule ASAP". We'll also modify the scheduler code to handle this and not try to schedule a second transfer for the same frame. Note that this change relies on the work to redo the microframe scheduler. It can work atop ("usb: dwc2: host: Manage frame nums better in scheduler") but it works even better after ("usb: dwc2: host: Totally redo the microframe scheduler"). With this change my stressful USB test (USB webcam + USB audio + keyboards) has less audio crackling than before. Acked-by:John Youn <johnyoun@synopsys.com> Signed-off-by:
Douglas Anderson <dianders@chromium.org> Tested-by:
Heiko Stuebner <heiko@sntech.de> Tested-by:
Stefan Wahren <stefan.wahren@i2se.com> Signed-off-by:
Felipe Balbi <balbi@kernel.org>
-
Douglas Anderson authored
As we start getting more exact about our scheduling it's becoming more and more important to know exactly how far through the current frame we are. This lets us make decisions about whether there's still time left to start a new transaction in the current frame. We'll add dwc2_hcd_get_future_frame_number() which will tell you what the frame number will be a certain number of microseconds (us) from now. We can use this information to help decide if there's enough time left in the frame for a transaction that will take a certain duration. This is expected to be used by a future change ("usb: dwc2: host: Properly set even/odd frame"). Acked-by: -
Douglas Anderson authored
The dwc2 scheduler (contained in hcd_queue.c) was a bit confusing in the way it initted / kept track of which frames a QH was going to be active in. Let's clean things up a little bit in preparation for a rewrite of the microframe scheduler. Specifically: * Old code would pick a frame number in dwc2_qh_init() and would try to pick it "in a slightly future (micro)frame". As far as I can tell the reason for this was that there was a delay between dwc2_qh_init() and when we actually wanted to dwc2_hcd_qh_add(). ...but apparently this attempt to be slightly in the future wasn't enough because dwc2_hcd_qh_add() then had code to reset things if the frame _wasn't_ in the future. There's no reason not to just pick the frame later. For non-periodic QH we now pick the frame in dwc2_hcd_qh_add(). For periodic QH we pick the frame at dwc2_schedule_periodic() time. * The old "dwc2_qh_init() actually assigned to "hsotg->frame_number". This doesn't seem like a great idea since that variable is supposed to be used to keep track of which SOF the interrupt handler has seen. Let's be clean: anyone who wants the current frame number (instead of the one as of the last interrupt) should ask for it. * The old code wasn't terribly consistent about trying to use the frame that the microframe scheduler assigned to it. In dwc2_sched_periodic_split() when it was scheduling the first frame it always "ORed" in 0x7 (!). Since the frame goes on the wire 1 uFrame after next_active_frame it meant that the SSPLIT would always try for uFrame 0 and the transaction would happen on the low speed bus during uFrame 1. This is irregardless of what the microframe scheduler said. * The old code assumed it would get called to schedule the next in a periodic split very quickly. That is if next_active_frame was 0 (transfer on wire in uFrame 1) it assumed it was getting called to schedule the next uFrame during uFrame 1 too (so it could queue something up for uFrame 2). It should be possible to actually queue something up for uFrame 2 while in uFrame 2 (AKA queue up ASAP). To do this, code needs to look at the previously scheduled frame when deciding when to next be active, not look at the current frame number. * If there was no microframe scheduler, the old code would check for whether we should be active using "qh->next_active_frame == frame_number". This seemed like a race waiting to happen. ...plus there's no way that you wouldn't want to schedule if next_active_frame was actually less than frame number. Note that this change doesn't make 100% sense on its own since it's expecting some sanity in the frame numbers assigned by the microframe scheduler and (as per the future patch which rewries it) I think that the current microframe scheduler is quite insane. However, it seems like splitting this up from the microframe scheduler patch makes things into smaller chunks and hopefully adds to clarity rather than reduces it. The two patches could certainly be squashed. Not that in the very least, I don't see any obvious bad behavior introduced with just this patch. I've attempted to keep the config parameter to disable the microframe scheduler in tact in this change, though I'm not sure it's worth it. Obviously the code is touched a lot so it's possible I regressed something when the microframe scheduler is disabled, though I did some basic testing and it seemed to work OK. I'm still not 100% sure why you wouldn't want the microframe scheduler (presuming it works), so maybe a future patch (or a future version of this patch?) could remove that parameter. Acked-by:
-
Douglas Anderson authored
We'll use the new "scheduler verbose debugging" macro to log missed SOFs. This is fast enough (assuming you configure it to use the ftrace buffer) that we can do it without worrying about the speed hit. The overhead hit if the scheduler tracing is set to "no_printk" should be near zero. Acked-by:
-
Douglas Anderson authored
This no-op change splits code out of dwc2_schedule_periodic() into a dwc2_do_reserve() function. This makes it a little easier to follow the logic. Acked-by:
-
Douglas Anderson authored
This no-op change just reorders a few functions in hcd_queue.c in order to prepare for future changes. Motivations here: The functions dwc2_hcd_qh_free() and dwc2_hcd_qh_create() are exported functions. They are not called within the file. That means that they should be near the bottom so that they can easily call static helpers. The function dwc2_qh_init() is only called by dwc2_hcd_qh_create() and should move near the bottom with it. The only reason that the dwc2_unreserve_timer_fn() timer function (and its subroutine dwc2_do_unreserve()) were so high in the file was that they needed to be above dwc2_qh_init(). Now that dwc2_qh_init() has been moved down it can be moved down a bit. A later patch will split the reserve code out of dwc2_schedule_periodic() and the reserve function should be near the unreserve function. The reserve function needs to be below dwc2_find_uframe() since it calls that. Acked-by:
-
Douglas Anderson authored
This no-op change just does some renames to simplify a future patch. 1. The "interval" field is renamed to "host_interval" to make it more obvious that this interval may be 8 times the interval that the device sees (if we're doing split transactions). A future patch will also add the "device_interval" field. 2. The "usecs" field is renamed to "host_us" again to make it more obvious that this is the time for the transaction as seen by the host. For split transactions the device may see a much longer transaction time. A future patch will also add "device_us". 3. The "sched_frame" field is renamed to "next_active_frame". The name "sched_frame" kept confusing me because it felt like something more permament (the QH's reservation or something). The name "next_active_frame" makes it more obvious that this field is constantly changing. Acked-by:
-
Douglas Anderson authored
The old code in dwc2_process_periodic_channels() would only enable the "periodic empty" interrupt if we weren't using DMA. That wasn't right since we can still get into cases where we have small FIFOs even on systems that have DMA (the rk3288 is a prime example). Let's always enable/disable the "periodic empty" when appropriate. As part of this: * Always call dwc2_process_periodic_channels() even if there's nothing in periodic_sched_assigned (we move the queue empty check so we still avoid the extra work). That will make extra certain that we will properly disable the "periodic empty" interrupt even if there's nothing queued up. * Move the enable of "periodic empty" due to non-empty periodic_sched_assigned to be for slave mode (non-DMA mode) only. Presumably this was the original intention of the check for DMA since it seems to match the comments above where in slave mode we leave things on the assigned queue. Note that even before this change slave mode didn't work for me, so I can't say for sure that my understanding of slave mode is correct. However, this shouldn't change anything for slave mode so if slave mode worked for someone in the past it ought to still work. With this change, I no longer get constant misses reported by my other debugging code (and with future patches) when I've got: * Rockchip rk3288 Chromebook, using port ff540000 -> Pluggable 7-port Hub with Charging (powered) -> Microsoft Wireless Keyboard 2000 in port 1. -> Das Keyboard in port 2. -> Jabra Speaker in port 3 -> Logitech, Inc. Webcam C600 in port 4 -> Microsoft Sidewinder X6 Keyboard in port 5 ...and I'm playing music on the USB speaker and capturing video from the webcam. Acked-by: -
Douglas Anderson authored
I find that when I plug a full speed (NOT high speed) hub into a dwc2 port and then I plug a bunch of devices into that full speed hub that dwc2 goes bat guano crazy. Specifically, it just spews errors like this in the console: usb usb1: clear tt 1 (9043) error -22 The specific test case I used looks like this: /: Bus 01.Port 1: Dev 1, Class=root_hub, Driver=dwc2/1p, 480M |__ Port 1: Dev 17, If 0, Class=Hub, Driver=hub/4p, 12M |__ Port 2: Dev 19, If 0, ..., Driver=usbhid, 1.5M |__ Port 4: Dev 20, If 0, ..., Driver=usbhid, 12M |__ Port 4: Dev 20, If 1, ..., Driver=usbhid, 12M |__ Port 4: Dev 20, If 2, ..., Driver=usbhid, 12M Showing VID/PID: Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub Bus 001 Device 017: ID 03eb:3301 Atmel Corp. at43301 4-Port Hub Bus 001 Device 020: ID 045e:0745 Microsoft Corp. Nano Transceiver ... Bus 001 Device 019: ID 046d:c404 Logitech, Inc. TrackMan Wheel I spent a bunch of time trying to figure out why there are errors to begin with. I believe that the issue may be a hardware issue where the transceiver sometimes accidentally sends a PREAMBLE packet if you send a packet to a full speed device right after one to a low speed device. Luckily the USB driver retries and the second time things work OK. In any case, things kinda seem work despite the errors, except for the "clear tt" spew mucking up my console. Chalk it up for a win for retries and robust protocols. So getting back to the "clear tt" problem, it appears that we get those because there's not actually a TT here to clear. It's my understanding that when dwc2 operates in low speed or full speed mode that there's no real TT out there. That makes all these attempts to "clear the TT" somewhat meaningless and also causes the spew in the log. Let's just skip all the useless TT clears. Eventually we should root cause the errors, but even if we do this is still a proper fix and is likely to avoid the "clear tt" error in the future. Note that hooking up a Full Speed USB Audio Device (Jabra 510) to this same hub with the keyboard / trackball shows that even audio works over this janky connection. As a point to note, this particular change (skip bogus TT clears) compared to just commenting out the dev_err() in hub_tt_work() actually produces better audio. Note: don't ask me where I got a full speed USB hub or whether the massive amount of dust that accumulated on it while it was in my junk box affected its funtionality. Just smile and nod. Acked-by:Kever Yang <kever.yang@rock-chips.com> Signed-off-by:
-
Douglas Anderson authored
According to the most up to date version of the dwc2 databook, the FRINT field of the HFIR register should be programmed to: * 125 us * (PHY clock freq for HS) - 1 * 1000 us * (PHY clock freq for FS/LS) - 1 This is opposed to older versions of the doc that claimed it should be: * 125 us * (PHY clock freq for HS) * 1000 us * (PHY clock freq for FS/LS) In case you didn't spot it, the difference is the "- 1". Let's add the "- 1" to match the newest user manual. It's presumed that the "- 1" should have always been there and that this was always a documentation error. If some hardware needs the "- 1" and other hardware doesn't, we'll have to add a configuration parameter for it in the future. I checked things before and after this patch on rk3288 using a Total Phase Beagle 5000 analyzer. Before this patch, a low speed mouse shows constant Frame Timing Jitter errors. After this patch errors have gone away. Before this patch SOF packets move forward about 1 us per 4 ms. After this patch the SOF packets move backward about 1 us per 255 ms. Some specific SOF timestamps from the analyzer are below. Before: 6.603.790 6.603.916 6.604.041 6.604.166 ... 6.607.541 6.607.667 6.607.792 6.607.917 ... 6.611.417 6.611.543 6.611.668 6.611.793 After: 6.215.159 6.215.284 6.215.408 6.215.533 6.215.658 ... 6.470.658 6.470.783 6.470.907 ... 6.726.032 6.726.157 6.725.281 6.725.406 Acked-by:
-
Douglas Anderson authored
In commit 94dfd7ed ("USB: HCD: support giveback of URB in tasklet context") support was added to give back the URB in tasklet context. Let's take advantage of this in dwc2. This speeds up the dwc2 interrupt handler considerably. Note that this requires the change ("usb: dwc2: host: Add a delay before releasing periodic bandwidth") to come first. Note that, as per Alan Stern in <https://patchwork.kernel.org/patch/7555771/>, we also need to make sure that the extra delay before the device drivers submit more data doesn't break the scheduler. At the moment the scheduler is pretty broken (see future patches) so it's hard to be 100% certain, but I have yet to see any new breakage introduced by this delay. ...and speeding up interrupt processing for dwc2 is a huge deal because it means we've got a better chance of not missing SOF interrupts. That means we've got an overall win here. Note that when playing USB audio and using a USB webcam and having several USB keyboards plugged in, the crackling on the USB audio device is noticably reduced with this patch. Acked-by:
-
Douglas Anderson authored
We'd like to be able to use HCD_BH in order to speed up the dwc2 host interrupt handler quite a bit. However, according to the kernel doc for usb_submit_urb() (specifically the part about "Reserved Bandwidth Transfers"), we need to keep a reservation active as long as a device driver keeps submitting. That was easy to do when we gave back the URB in the interrupt context: we just looked at when our queue was empty and released the reserved bandwidth then. ...but now we need a little more complexity. We'll follow EHCI's lead in commit 9118f9eb ("USB: EHCI: improve interrupt qh unlink") and add a 5ms delay. Since we don't have a whole timer infrastructure in dwc2, we'll just add a timer per QH. The overhead for this is very small. Note that the dwc2 scheduler is pretty broken (see future patches to fix it). This patch attempts to replicate all old behavior and just add the proper delay. Acked-by:
-
Douglas Anderson authored
In preparation for future changes to the scheduler let's add some tracing that makes it easy for us to see what's happening. By default this tracing will be off. By changing "core.h" you can easily trace to ftrace, the console, or nowhere. Acked-by:
-
Douglas Anderson authored
We're supposed to keep outstanding splits in order. Keep track of a list of the order of splits and process channel interrupts in that order. Without this change and the following setup: * Rockchip rk3288 Chromebook, using port ff540000 -> Pluggable 7-port Hub with Charging (powered) -> Microsoft Wireless Keyboard 2000 in port 1. -> Das Keyboard in port 2. ...I find that I get dropped keys on the Microsoft keyboard (I'm sure there are other combinations that fail, but this documents my test). Specifically I've been typing "hahahahahahaha" on the keyboard and often see keys dropped or repeated. After this change the above setup works properly. This patch is based on a previous patch proposed by Yunzhi Li ("usb: dwc2: hcd: fix periodic transfer schedule sequence") Acked-by:Yunzhi Li <lyz@rock-chips.com> Reviewed-by:
-
Douglas Anderson authored
The queues the the dwc2 host controller used are truly queues. That means FIFO or first in first out. Unfortunately though the code was iterating through these queues starting from the head, some places in the code was adding things to the queue by adding at the head instead of the tail. That means last in first out. Doh. Go through and just always add to the tail. Doing this makes things much happier when I've got: * 7-port USB 2.0 Single-TT hub * - Microsoft 2.4 GHz Transceiver v7.0 dongle * - Jabra speakerphone playing music Acked-by:
-
Douglas Anderson authored
When poking around with USB devices with slub_debug enabled, I found another obvious use after free. Turns out that in dwc2_hc_n_intr() I was in a state when the contents of chan->qh was filled with 0x6b, indicating that chan->qh was freed but chan still had a reference to it. Let's make sure that whenever we free qh we also make sure we remove a reference from its channel. The bug fixed here doesn't appear to be new--I believe I just got lucky and happened to see it while stress testing. Acked-by:
-
Douglas Anderson authored
As documented in dwc2_calculate_dynamic_fifo(), host_rx_fifo_size should really be: 2 * ((Largest Packet size / 4) + 1 + 1) + n with n = number of host channel. We have 9 host channels, so 2 * ((1024/4) + 2) + 9 = 516 + 9 = 525 We've got 960 / 972 total_fifo_size on rk3288 (and presumably on rk3066) and 525 + 128 + 256 = 909 so we're still under on both ports even when we increment by 5. In the future, it would be nice if dwc2_calculate_dynamic_fifo() could handle the "too small" FIFO case and come up with something more dynamically. When we do that we can figure out how to allocate the extra 48 / 60 bytes of FIFO that we're currently wasting. NOTE: no known bugs are fixed by this patch, but it seems like a simple fix and ought to fix someone. Acked-by:
-
Douglas Anderson authored
All other host controllers who want aligned buffers for DMA do it a certain way. Let's do that too instead of working behind the USB core's back. This makes our interrupt handler not take forever and also rips out a lot of code, simplifying things a bunch. This also has the side effect of removing the 65535 max transfer size limit. NOTE: The actual code to allocate the aligned buffers is ripped almost completely from the tegra EHCI driver. At some point in the future we may want to add this functionality to the USB core to share more code everywhere. Signed-off-by:
-
Douglas Anderson authored
Previously we needed to set the max_transfer_size to explicitly be 65535 because the old driver would detect that our hardware could support much bigger transfers and then would try to do them. This wouldn't work since the DMA alignment code couldn't support it. Later in commit e8f8c14d ("usb: dwc2: clip max_transfer_size to 65535") upstream added support for clipping this automatically. Since that commit it has been OK to just use "-1" (default), but nobody bothered to change it. Let's change it to default now for two reasons: - It's nice to use autodetected params. - If we can remove the 65535 limit, we can transfer more! Signed-off-by:
-
John Youn authored
Check that dwc->maximum_speed is set to a valid value. Also add an error when we use it later if we encounter an invalid value. Signed-off-by:
-
Emilio López authored
MODULE_DEVICE_TABLE() is missing, so the module isn't auto-loading on sunxi systems using the OTG controller. This commit adds the missing line so it loads automatically when building it as a module and running on a system with an USB OTG port. Signed-off-by:
Emilio López <emilio.lopez@collabora.co.uk> Signed-off-by:
-
Li Jun authored
B-device detects that bus is idle for more than TB_AIDL_BDIS min and begins HNP by turning off pullup on DP, this allows the bus to discharge to the SE0 state. This timer was missed and failed with PET test: 6.8.5 B-UUT HNP of USB OTG and EH automated compliance plan v1.2, this patch is to fix this timing issue. Acked-by:
Peter Chen <peter.chen@nxp.com> Signed-off-by:
Li Jun <jun.li@nxp.com> Signed-off-by:
-
Li Jun authored
Add A-idle to B-disconnect timer, B-device detects that bus is idle for more than TB_AIDL_BDIS min and begins HNP by turning off pullup on D+. This allows the bus to discharge to the SE0 state. Acked-by:
-
Li Jun authored
Update HNP test procedure as HNP polling is supported. Acked-by:
-
Li Jun authored
Enable HNP polling support for chipidea gadget and allocate memory for host request flag when otg fsm init. Acked-by:
-
Li Jun authored
Set host_request_flag if the current peripheral wants to take host role via changing a_bus_req or b_bus_req by user application. Acked-by:
Peter Chen <peter.chen@freescale.com> Signed-off-by:
-
Li Jun authored
If gadget with HNP polling support receives GetStatus request of otg status selector, it feedback to host with host request flag to indicate if it wants to take host role. Acked-by:
-
Li Jun authored
Since gadget driver will handle this request, so controller driver bypass it. Acked-by:
-
Li Jun authored
Adds HNP polling timer when transits to host state, the OTG status request will be sent to peripheral after timeout, if host request flag is set, it will switch to peripheral state, otherwise it will repeat HNP polling every 1.5s and maintain the current session. Acked-by:
-
Li Jun authored
A host is required to use the GetStatus command, with wIndex set to the OTG status selector(F000H) to request the Host request flag from the peripheral. Acked-by:
-
Li Jun authored
Add 2 flags for USB OTG HNP polling, hnp_polling_support is to indicate if the gadget can support HNP polling, host_request_flag is used for gadget to store host request information from application, which can be used to respond to HNP polling from host. Acked-by:
-
Marek Szyprowski authored
Since commit 855ed04a ("usb: gadget: udc-core: independent registration of gadgets and gadget drivers") gadget drivers can not assume that UDC drivers are already available on their initialization. This broke the HACK, which was used in gadgetfs driver, to get UDC controller name. This patch removes this hack and replaces it by additional function in the UDC core (which is usefully only for legacy drivers, please don't use it in the new code). Reported-by:
Vegard Nossum <vegard.nossum@oracle.com> Signed-off-by:
Marek Szyprowski <m.szyprowski@samsung.com> Tested-by:
-
Michal Nazarewicz authored
For every in_substream, there must be a corresponding gmidi_in_port structure so it is perfectly viable and some might argue sensible to stash pointer to the input substream in the gmidi_in_port structure. This has an added benefit that if in_ports < MAX_PORTS, the whole f_midi structure takes up less space because only in_ports number of pointers for in_substream are allocated instead of MAX_PORTS lots of them. Signed-off-by:
Michal Nazarewicz <mina86@mina86.com> Signed-off-by:
-
Dan Carpenter authored
We added a new error path to this function and we forgot to drop the lock. Fixes: e1e3d7ec ('usb: gadget: f_midi: pre-allocate IN requests') Signed-off-by:
Dan Carpenter <dan.carpenter@oracle.com> [mina86@mina86.com: rebased on top of refactoring commit] Signed-off-by:
-
Michal Nazarewicz authored
Reduce number of allocations, simplify memory management and reduce memory usage by stacking the gmidi_in_port elements at the end of the f_midi structure using a flexible array. Signed-off-by:
-
Michal Nazarewicz authored
In general case, all of midi->in_port pointers may be non-NULL which implies that the ‘if (\!port)’ condition will never execute thus never zeroing midi->in_last_port. Fix by rewriting the loop such that the field is set to zero if \!port or end of loop has been reached. Signed-off-by:
-
Michal Nazarewicz authored
Move some of the f_midi_transmit to a separate f_midi_do_transmit function so the massive indention levels are not so jarring. This introduces no changes in behaviour. Signed-off-by:
-
Felipe F. Tonello authored
remove a field which is unnecessary. No functional changes. Signed-off-by:
Felipe F. Tonello <eu@felipetonello.com> Signed-off-by:
-
Du, Changbin authored
ffs_epfile_io and ffs_epfile_io_complete runs in different context, but there is no synchronization between them. consider the following scenario: 1) ffs_epfile_io interrupted by sigal while wait_for_completion_interruptible 2) then ffs_epfile_io set ret to -EINTR 3) just before or during usb_ep_dequeue, the request completed 4) ffs_epfile_io return with -EINTR In this case, ffs_epfile_io tell caller no transfer success but actually it may has been done. This break the caller's pipe. Below script can help test it (adbd is the process which lies on f_fs). while true do pkill -19 adbd #SIGSTOP pkill -18 adbd #SIGCONT sleep 0.1 done To avoid this, just dequeue the request first. After usb_ep_dequeue, the request must be done or canceled. With this change, we can ensure no race condition in f_fs driver. But actually I found some of the udc driver has analogical issue in its dequeue implementation. For example, 1) the dequeue function hold the controller's lock. 2) before driver request controller to stop transfer, a request completed. 3) the controller trigger a interrupt, but its irq handler need wait dequeue function to release the lock. 4) dequeue function give back the request with negative status, and release lock. 5) irq handler get lock but the request has already been given back. So, the dequeue implementation should take care of this case. IMO, it can be done as below steps to dequeue a already started request, 1) request controller to stop transfer on the given ep. HW know the actual transfer status. 2) after hw stop transfer, driver scan if there are any completed one. 3) if found, process it with real status. if no, the request can canceled. Signed-off-by:
"Du, Changbin" <changbin.du@intel.com> [mina86@mina86.com: rebased on top of refactoring commits] Signed-off-by:
-
Michal Nazarewicz authored
Eliminate one of the return paths by using a ‘goto error_mutex’ and rearrange some if-bodies which results in reduction of the indention level and thus hopefully makes the function easier to read and reason about. Signed-off-by:
-
