- 12 Sep, 2016 1 commit
-
-
Lu Baolu authored
Commit 50c763f8 ("usb: dwc3: Set the ClearPendIN bit on Clear Stall EP command") sets ClearPendIN bit for all IN endpoints of v2.60a+ cores. This causes ClearStall command fails on 2.60+ cores operating in HighSpeed mode. In page 539 of 2.60a specification: "When issuing Clear Stall command for IN endpoints in SuperSpeed mode, the software must set the "ClearPendIN" bit to '1' to clear any pending IN transcations, so that the device does not expect any ACK TP from the host for the data sent earlier." It's obvious that we only need to apply this rule to those IN endpoints that currently operating in SuperSpeed mode. Fixes: 50c763f8 ("usb: dwc3: Set the ClearPendIN bit on Clear Stall EP command") Cc: <stable@vger.kernel.org> # v4.7+ Signed-off-by:
Lu Baolu <baolu.lu@linux.intel.com> Signed-off-by:
Felipe Balbi <felipe.balbi@linux.intel.com>
-
- 09 Sep, 2016 4 commits
-
-
Arnd Bergmann authored
Cleaning up the loop in dwc3_cleanup_done_reqs() introduced a gcc warning if built with "-Wmaybe-uninitialized": drivers/usb/dwc3/gadget.c: In function 'dwc3_endpoint_transfer_complete': drivers/usb/dwc3/gadget.c:2015:9: 'trb' may be used uninitialized in this function [-Wmaybe-uninitialized] I believe it is a false positive and we always have a valid 'trb' pointer at the end of the function, but neither I nor the compiler are able to prove that. This works around the warning by computing a flag earlier in the function when it's guaranteed to be valid, which tells the compiler that it's safe and makes it easier to understand to me. Signed-off-by:
Arnd Bergmann <arnd@arndb.de> Fixes: 31162af4 ("usb: dwc3: gadget: avoid while (1) loop on completion") Signed-off-by:
-
Harish Jenny K N authored
This commit incorporates findings from https://lkml.org/lkml/2016/4/25/594 The function has been modified to make sure we hold the dev lock when accessing the net device pointer. Acked-by:
Jim Baxter <jim_baxter@mentor.com> Signed-off-by:
Harish Jenny K N <harish_kandiga@mentor.com> Signed-off-by:
-
Harish Jenny K N authored
dev->port_usb is checked for null pointer previously, so dev->port_usb might be null during no zlp check, fix it by adding null pointer check. Acked-by:
-
Colin Ian King authored
the if statement in lb_modinit is unnecessary so we can totally remove the variable ret and just return the return value from the call to usb_function_register. Signed-off-by:
Colin Ian King <colin.king@canonical.com> Signed-off-by:
-
- 08 Sep, 2016 4 commits
-
-
John Youn authored
When a force mode bit is set and the IDDIG debounce filter is enabled, there is a delay for the forced mode to take effect. This delay is due to the IDDIG debounce filter and is variable depending on the platform's PHY clock speed. To account for this delay we can poll for the expected mode. On a clear force mode, since we don't know what mode to poll for, delay for a fixed 100 ms. This is the maximum delay based on the slowest PHY clock speed. Tested-by:
Stefan Wahren <stefan.wahren@i2se.com> Signed-off-by:
John Youn <johnyoun@synopsys.com> Signed-off-by:
-
John Youn authored
Add a delay to the core soft reset function to account for the IDDIG debounce filter. If the current mode is host, either due to the force mode bit being set (which persists after core reset) or the connector id pin, a core soft reset will temporarily reset the mode to device and a delay from the IDDIG debounce filter will occur before going back to host mode. Tested-by:
-
John Youn authored
In dwc2_hsotg_udc_start(), don't initialize the controller for device mode unless we are actually in device mode. Tested-by:
Heiko Stuebner <heiko@sntech.de> Tested-by:
-
Jim Baxter authored
The struct ffs_data::private_data has a pointer to ffs_dev stored in it during the ffs_fs_mount() function however it is not cleared when the ffs_dev is freed later which causes the ffs_closed function to crash with "Unable to handle kernel NULL pointer dereference" error when using the data in ffs_data::private_data. This clears this pointer during the ffs_free_dev clean up function. Signed-off-by:
Jiada Wang <jiada_wang@mentor.com> Signed-off-by:
-
- 07 Sep, 2016 1 commit
-
-
Thinh Nguyen authored
This patch follows the similar fix in dwc2. See commit 5268ed9d ("usb: dwc2: Fix dr_mode validation") Currently, the dr_mode is only checked against the module configuration. It also needs to be checked against the hardware capablities. The driver now checks if both the module configuration and hardware are capable of the dr_mode value. If not, then it will issue a warning and fall back to a supported value. If it is unable to fall back to a suitable value, then the probe will fail. Behavior summary: module : actual HW config dr_mode : dr_mode --------------------------------- host host any : host host dev any : INVALID host otg any : host dev host any : INVALID dev dev any : dev dev otg any : dev otg host any : host otg dev any : dev otg otg any : dr_mode Signed-off-by:
Thinh Nguyen <thinhn@synopsys.com> Signed-off-by:
-
- 06 Sep, 2016 9 commits
-
-
Tal Shorer authored
ulpi_register_interface() accepts a const struct ulpi_ops and dwc3 doesn't perform any changes to this struct at runtime, so there's no reason it shouldn't be constant. Reviewed-by:
Heikki Krogerus <heikki.krogerus@linux.intel.com> Signed-off-by:
Tal Shorer <tal.shorer@gmail.com> Signed-off-by:
-
Tal Shorer authored
None of the core ulpi functions perform any changes to the operations struct, and logically as a struct that contains function pointers there's no reason it shouldn't be constant. Acked-by:
-
Tal Shorer authored
Operations now use ulpi->dev.parent directly instead of via the ulpi_ops struct, making this field unused. Remove it. Acked-by:
-
Tal Shorer authored
With the removal of the old {read|write} operations, we can now safely rename the new api operations {read|write}_dev to use the shorter and clearer names {read|write}, respectively. Acked-by: -
Tal Shorer authored
Now that all users use the new api callbacks, remove the old api callbacks and force new interface drivers to use the new api. Acked-by:
-
Tal Shorer authored
The old read, write callbacks in struct ulpi_ops have been deprecated in favor of new callbacks that pass the parent device directly. Replace the used callbacks in dwc3's ulpi component with the new api. Reviewed-by:
-
Tal Shorer authored
Add these two new api callbacks to struct ulpi_ops. These are different than read, write in that they pass the parent device directly instead of via the ops argument. They are intended to replace the old api functions. If the new api callbacks are missing, revert to calling the old ones as before. Acked-by:
-
Tal Shorer authored
Once ulpi operations use the parent device directly, this will be needed during the operations used in ulpi_register() itself, so set the parent field before calling any ulpi operations. Acked-by:
-
John Youn authored
This property is not needed because the periodic fifos are not configurable. So it was incorrect to add this property in the first place. Acked-by:
Rob Herring <robh@kernel.org> Signed-off-by:
-
- 31 Aug, 2016 13 commits
-
-
Robert Baldyga authored
As kill_all_requests() potentially flushes TX FIFO, we should should free FIFO after calling it. Otherwise FIFO could stay unflushed properly. Signed-off-by:
Robert Baldyga <r.baldyga@samsung.com> Signed-off-by:
-
Robert Baldyga authored
Since FIFO is always freed in dwc2_hsotg_ep_disable(), ep->fifo_index is always 0 in dwc2_hsotg_ep_enable(), hence code inside if() block is never executed. Signed-off-by:
-
Robert Baldyga authored
Since we handle FIFOs and endpoint separately, using variable named 'ep' in context of FIFO is misleading, hence we rename it to 'fifo'. Signed-off-by:
-
Robert Baldyga authored
According to DWC2 documentation, DPTxFSize field of DPTXFSIZn register is read only, which means that software cannot change FIFO size. Register description says: "The value of this register is the Largest Device Mode Periodic Tx Data FIFO Depth (parameter OTG_TX_DPERIO_DFIFO_DEPTH_n), as specified during coreConsultant configuration." That means, that we have to setup only FIFO start addresses (DPTxFStAddr), taking into account reset values of DPTxFSize. Initialize FIFO start addresses properly and remove unneeded core related to incorrect FIFO size initialization. Signed-off-by:
-
Robert Baldyga authored
In context of FIFO registers we use ep->fifo_index instead of ep->index. Signed-off-by:
-
Romain Izard authored
Disabling USB gadget functions configured through configfs is something that can happen in normal use cases. Keep the existing log for this type of event, but only as debug, not as an error. Signed-off-by:
Romain Izard <romain.izard.pro@gmail.com> Signed-off-by:
-
Petr Cvek authored
Some UVC commands require additional data (non zero uvc->event_length). Add usb_ep_queue() call, so uvc_function_ep0_complete() can be called and send received data to the userspace. Signed-off-by:
Petr Cvek <petr.cvek@tul.cz> Signed-off-by:
-
Yoshihiro Shimoda authored
This patch sets the quirk_avoids_skb_reserve flag to improve performance of a network gadget driver (e.g. f_ncm.c) if USB-DMAC is used. For example (on r8a7795 board + f_ncm.c + iperf udp mode / receiving): - without this patch: 90.3 Mbits/sec - with this patch: 273 Mbits/sec Signed-off-by:
Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com> Signed-off-by:
-
Yoshihiro Shimoda authored
This patch adds to support no_skb_reserve function to improve performance for some platforms. About the detail, please refer to the commit log of "quirk_avoids_skb_reserve" in include/linux/usb/gadget.h. Signed-off-by:
-
Yoshihiro Shimoda authored
This patch adds a flag "no_skb_reserve" in struct eth_dev. So, if a peripheral driver sets the quirk_avoids_skb_reserve flag, upper network gadget drivers (e.g. f_ncm.c) can avoid skb_reserve() calling using the flag as well. Signed-off-by:
-
Yoshihiro Shimoda authored
Some platforms (e.g. USB-DMAC on R-Car SoCs) has memory alignment restriction. If memory alignment is not match, the usb peripheral driver decides not to use the DMA controller. Then, the performance is not good. In the case of u_ether.c, since it calls skb_reserve() in rx_submit(), it is possible to cause memory alignment mismatch. So, this patch adds a new quirk "quirk_avoids_skb_reserve" to avoid skb_reserve() calling in u_ether.c to improve performance. A peripheral driver will set this flag and network gadget drivers (e.g. f_ncm.c) will reference the flag via gadget_avoids_skb_reserve(). Signed-off-by:
-
Colin Ian King authored
trivial typo fix in dev_err message Signed-off-by:
-
Yoshihiro Shimoda authored
This patch adds support for r8a7796 (R-Car M3-W). Acked-by:
Geert Uytterhoeven <geert+renesas@glider.be> Acked-by:
-
- 29 Aug, 2016 8 commits
-
-
Alexey Khoroshilov authored
Memory allocated for goku_udc device is not deallocated at error paths in goku_probe(), because gadget_release() destructor is not registered yet. Found by Linux Driver Verification project (linuxtesting.org). Signed-off-by:
Alexey Khoroshilov <khoroshilov@ispras.ru> Signed-off-by:
-
Wolfram Sang authored
All kmalloc-based functions print enough information on failures. Signed-off-by:
Wolfram Sang <wsa-dev@sang-engineering.com> Signed-off-by:
-
Wolfram Sang authored
All kmalloc-based functions print enough information on failures. Signed-off-by:
-
Wolfram Sang authored
All kmalloc-based functions print enough information on failures. Signed-off-by:
-
Wolfram Sang authored
All kmalloc-based functions print enough information on failures. Signed-off-by:
-
Wolfram Sang authored
All kmalloc-based functions print enough information on failures. Signed-off-by:
-
Wolfram Sang authored
All kmalloc-based functions print enough information on failures. Signed-off-by:
-
Wolfram Sang authored
All kmalloc-based functions print enough information on failures. Signed-off-by:
-
