Add logic to consolidate TRBs for Synopsys xHC
Subsystems (3)
v4 cover Add logic to consolidate TRBs for Synopsys xHC
The Synopsys xHC has an internal TRB cache of size TRB_CACHE_SIZE for each endpoint. The default value for TRB_CACHE_SIZE is 16 for SS and 8 for HS. The controller loads and updates the TRB cache from the transfer ring in system memory whenever the driver issues a start transfer or update transfer command.
For chained TRBs, the Synopsys xHC requires that the total amount of bytes for all TRBs loaded in the TRB cache be greater than or equal to 1 MPS. Or the chain ends within the TRB cache (with a last TRB).
If this requirement is not met, the controller will not be able to send or receive a packet and it will hang causing a driver timeout and error.
This patch set adds logic to the XHCI driver to detect and prevent this from happening along with the quirk to enable this logic for Synopsys HAPS platform.
Based on Mathias's feedback on previous implementation where consolidation was done in TRB cache, with this patch series the implementation is done during mapping of the URB by consolidating the SG list into a temporary buffer if the SG list buffer sizes within TRB_CACHE_SIZE is less than MPS.
Changes since v3:
- Removed the dt-binding patch
- Added new patch to pass the quirk as platform data
- Modified the patch to set the quirk
Changes since v2:
- Modified the xhci_unmap_temp_buffer function to unmap dma and c the dma flag
- Removed RFC tag
Changes since v1:
- Comments from Greg are addressed on [PATCH 4/4] and [PATCH 1/4]
- Renamed the property and quirk as in other patches based on [PATCH 1/4]
Tejas Joglekar (3):
usb: xhci: Set quirk for XHCI_SG_TRB_CACHE_SIZE_QUIRK
usb: xhci: Use temporary buffer to consolidate SG
usb: dwc3: Pass quirk as platform data
drivers/usb/dwc3/host.c | 14 ++++ drivers/usb/host/xhci-plat.c | 3 + drivers/usb/host/xhci-ring.c | 2 +- drivers/usb/host/xhci.c | 137 ++++++++++++++++++++++++++++++++++- drivers/usb/host/xhci.h | 5 ++ 5 files changed, 159 insertions(+), 2 deletions(-)
Signature · 1 line
2.28.0
PATCH v4 1/3 usb: xhci: Set quirk for XHCI_SG_TRB_CACHE_SIZE_QUIRK
This commit uses the private data passed by parent device to set the quirk for Synopsys xHC. This patch fixes the SNPS xHC hang issue when the data is scattered across small buffers which does not make atleast MPS size for given TRB cache size of SNPS xHC.
--- drivers/usb/host/xhci-plat.c | 3 +++ drivers/usb/host/xhci.h | 1 + 2 files changed, 4 insertions(+)
drivers/usb/host/xhci-plat.c +3 -0
@@ -333,6 +333,9 @@ static int xhci_plat_probe(struct platform_device *pdev)if (priv && (priv->quirks &XHCI_SKIP_PHY_INIT )) hcd->skip_phy_initialization =1 ; +if (priv && (priv->quirks &XHCI_SG_TRB_CACHE_SIZE_QUIRK )) + xhci->quirks |=XHCI_SG_TRB_CACHE_SIZE_QUIRK ; + ret =usb_add_hcd (hcd, irq,IRQF_SHARED );if (ret) goto disable_usb_phy;
drivers/usb/host/xhci.h +1 -0
@@ -1877,6 +1877,7 @@ struct xhci_hcd {#define XHCI_SNPS_BROKEN_SUSPEND BIT_ULL(35)#define XHCI_RENESAS_FW_QUIRK BIT_ULL(36)#define XHCI_SKIP_PHY_INIT BIT_ULL(37) +#define XHCI_SG_TRB_CACHE_SIZE_QUIRK BIT_ULL(38)unsigned int num_active_eps;unsigned int limit_active_eps;
Signature · 1 line
2.28.0
PATCH v4 2/3 usb: xhci: Use temporary buffer to consolidate SG
In reply to Tejas JoglekarThe Synopsys xHC has an internal TRB cache of size TRB_CACHE_SIZE for each endpoint. The default value for TRB_CACHE_SIZE is 16 for SS and 8 for HS. The controller loads and updates the TRB cache from the transfer ring in system memory whenever the driver issues a start transfer or update transfer command.
For chained TRBs, the Synopsys xHC requires that the total amount of bytes for all TRBs loaded in the TRB cache be greater than or equal to 1 MPS. Or the chain ends within the TRB cache (with a last TRB).
If this requirement is not met, the controller will not be able to send or receive a packet and it will hang causing a driver timeout and error.
This can be a problem if a class driver queues SG requests with many small-buffer entries. The XHCI driver will create a chained TRB for each entry which may trigger this issue.
This patch adds logic to the XHCI driver to detect and prevent this from happening.
For every (TRB_CACHE_SIZE - 2), we check the total buffer size of the SG list and if the last window of (TRB_CACHE_SIZE - 2) SG list length and we don't make up at least 1 MPS, we create a temporary buffer to consolidate full SG list into the buffer.
We check at (TRB_CACHE_SIZE - 2) window because it is possible that there would be a link and/or event data TRB that take up to 2 of the cache entries.
We discovered this issue with devices on other platforms but have not yet come across any device that triggers this on Linux. But it could be a real problem now or in the future. All it takes is N number of small chained TRBs. And other instances of the Synopsys IP may have smaller values for the TRB_CACHE_SIZE which would exacerbate the problem.
--- drivers/usb/host/xhci-ring.c | 2 +- drivers/usb/host/xhci.c | 137 ++++++++++++++++++++++++++++++++++- drivers/usb/host/xhci.h | 4 + 3 files changed, 141 insertions(+), 2 deletions(-)
drivers/usb/host/xhci-ring.c +1 -1
@@ -3325,7 +3325,7 @@ int xhci_queue_bulk_tx(struct xhci_hcd *xhci, gfp_t mem_flags, full_len = urb->transfer_buffer_length;/* If we have scatter/gather list, we use it. */ -if (urb->num_sgs) { +if (urb->num_sgs && !(urb->transfer_flags &URB_DMA_MAP_SINGLE )) { num_sgs = urb->num_mapped_sgs; sg = urb->sg; addr = (u64 )sg_dma_address (sg);
drivers/usb/host/xhci.c +136 -1
161 lines not on this page · raw message
drivers/usb/host/xhci.h +4 -0
@@ -1330,6 +1330,10 @@ enum xhci_setup_dev {#define TRB_SIA (1<<31)#define TRB_FRAME_ID (p) (((p) & 0x7ff) << 20) +/* TRB cache size for xHC with TRB cache */ +#define TRB_CACHE_SIZE_HS 8 +#define TRB_CACHE_SIZE_SS 16 +struct xhci_generic_trb {__le32 field[4 ]; };
Signature · 1 line
2.28.0
kernel test robot
build botRe: [PATCH v4 2/3] usb: xhci: Use temporary buffer to consolidate SG
Thank you for the patch! Perhaps something to improve:
Hi Tejas,
Thank you for the patch! Perhaps something to improve:
[auto build test WARNING on usb/usb-testing]
[also build test WARNING on next-20201013]
[cannot apply to balbi-usb/testing/next peter.chen-usb/ci-for-usb-next v5.9] [If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]
url: https://github.com/0day-ci/linux/commits/Tejas-Joglekar/Add-logic-to-consolidate-TRBs-for-Synopsys-xHC/20201013-204605 base: https://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb.git usb-testing config: m68k-randconfig-m031-20201013 (attached as .config) compiler: m68k-linux-gcc (GCC) 9.3.0
reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# https://github.com/0day-ci/linux/commit/9d7453725e3adcd62b021dbf4cbd4f29476e71e4
git remote add linux-review https://github.com/0day-ci/linux
git fetch --no-tags linux-review Tejas-Joglekar/Add-logic-to-consolidate-TRBs-for-Synopsys-xHC/20201013-204605 git checkout 9d7453725e3adcd62b021dbf4cbd4f29476e71e4
# save the attached .config to linux build tree
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-9.3.0 make.cross ARCH=m68k
If you fix the issue, kindly add following tag as appropriate
All warnings (new ones prefixed by >>):
drivers/usb/host/xhci.c: In function 'xhci_map_temp_buffer':
>> drivers/usb/host/xhci.c:1269:19: warning: variable 'xhci' set but not used [-Wunused-but-set-variable]
1269 | struct xhci_hcd *xhci;
| ^~~~
>> drivers/usb/host/xhci.c:1266:15: warning: variable 'len' set but not used [-Wunused-but-set-variable]
1266 | unsigned int len;
| ^~~
drivers/usb/host/xhci.c: In function 'xhci_urb_temp_buffer_required':
>> drivers/usb/host/xhci.c:1305:15: warning: variable 'buf_len' set but not used [-Wunused-but-set-variable]
1305 | unsigned int buf_len;
| ^~~~~~~
drivers/usb/host/xhci.c: In function 'xhci_unmap_temp_buf':
drivers/usb/host/xhci.c:1345:15: warning: variable 'len' set but not used [-Wunused-but-set-variable]
1345 | unsigned int len;
| ^~~
>> drivers/usb/host/xhci.c:1344:22: warning: variable 'sg' set but not used [-Wunused-but-set-variable]
1344 | struct scatterlist *sg;
| ^~vim +/xhci +1269 drivers/usb/host/xhci.c
1261
1262 static int xhci_map_temp_buffer(struct usb_hcd *hcd, struct urb *urb)
1263 {
1264 void *temp;
1265 int ret = 0;
> 1266 unsigned int len;
1267 unsigned int buf_len;
1268 enum dma_data_direction dir;
> 1269 struct xhci_hcd *xhci;
1270
1271 xhci = hcd_to_xhci(hcd);
1272 dir = usb_urb_dir_in(urb) ? DMA_FROM_DEVICE : DMA_TO_DEVICE;
1273 buf_len = urb->transfer_buffer_length;
1274
1275 temp = kzalloc_node(buf_len, GFP_ATOMIC,
1276 dev_to_node(hcd->self.sysdev));
1277
1278 if (usb_urb_dir_out(urb))
1279 len = sg_pcopy_to_buffer(urb->sg, urb->num_sgs,
1280 temp, buf_len, 0);
1281
1282 urb->transfer_buffer = temp;
1283 urb->transfer_dma = dma_map_single(hcd->self.sysdev,
1284 urb->transfer_buffer,
1285 urb->transfer_buffer_length,
1286 dir);
1287
1288 if (dma_mapping_error(hcd->self.sysdev,
1289 urb->transfer_dma)) {
1290 ret = -EAGAIN;
1291 kfree(temp);
1292 } else {
1293 urb->transfer_flags |= URB_DMA_MAP_SINGLE;
1294 }
1295
1296 return ret;
1297 }
1298
1299 static bool xhci_urb_temp_buffer_required(struct usb_hcd *hcd,
1300 struct urb *urb)
1301 {
1302 bool ret = false;
1303 unsigned int i;
1304 unsigned int len = 0;
> 1305 unsigned int buf_len;
1306 unsigned int trb_size;
1307 unsigned int max_pkt;
1308 struct scatterlist *sg;
1309 struct scatterlist *tail_sg;
1310
1311 sg = urb->sg;
1312 tail_sg = urb->sg;
1313 buf_len = urb->transfer_buffer_length;
1314 max_pkt = usb_endpoint_maxp(&urb->ep->desc);
1315
1316 if (!urb->num_sgs)
1317 return ret;
1318
1319 if (urb->dev->speed >= USB_SPEED_SUPER)
1320 trb_size = TRB_CACHE_SIZE_SS;
1321 else
1322 trb_size = TRB_CACHE_SIZE_HS;
1323
1324 if (urb->transfer_buffer_length != 0 &&
1325 !(urb->transfer_flags & URB_NO_TRANSFER_DMA_MAP)) {
1326 for_each_sg(urb->sg, sg, urb->num_sgs, i) {
1327 len = len + sg->length;
1328 if (i > trb_size - 2) {
1329 len = len - tail_sg->length;
1330 if (len < max_pkt) {
1331 ret = true;
1332 break;
1333 }
1334
1335 tail_sg = sg_next(tail_sg);
1336 }
1337 }
1338 }
1339 return ret;
1340 }
1341
1342 static void xhci_unmap_temp_buf(struct usb_hcd *hcd, struct urb *urb)
1343 {
> 1344 struct scatterlist *sg;
1345 unsigned int len;
1346 unsigned int buf_len;
1347 enum dma_data_direction dir;
1348
1349 dir = usb_urb_dir_in(urb) ? DMA_FROM_DEVICE : DMA_TO_DEVICE;
1350
1351 sg = urb->sg;
1352 buf_len = urb->transfer_buffer_length;
1353
1354 if (IS_ENABLED(CONFIG_HAS_DMA) &&
1355 (urb->transfer_flags & URB_DMA_MAP_SINGLE))
1356 dma_unmap_single(hcd->self.sysdev,
1357 urb->transfer_dma,
1358 urb->transfer_buffer_length,
1359 dir);
1360
1361 if (usb_urb_dir_in(urb))
1362 len = sg_pcopy_from_buffer(urb->sg, urb->num_sgs,
1363 urb->transfer_buffer,
1364 buf_len,
1365 0); 1366
1367 urb->transfer_flags &= ~URB_DMA_MAP_SINGLE;
1368 kfree(urb->transfer_buffer);
1369 urb->transfer_buffer = NULL;
1370 }
1371
---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org
PATCH v4 3/3 usb: dwc3: Pass quirk as platform data
In reply to Tejas JoglekarThis commit adds the platform device data to setup the XHCI_SG_TRB_CACHE_SIZE_QUIRK quirk. DWC3 hosts which are PCI devices does not use OF to create platform device but create xhci-plat platform device runtime. So
this patch allow parent device to supply the quirk through platform data.
--- drivers/usb/dwc3/host.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+)
drivers/usb/dwc3/host.c +14 -0
@@ -11,6 +11,15 @@#include <linux/platform_device.h> #include "core.h" +#include "../host/xhci-plat.h" + +static const struct xhci_plat_priv dwc3_pdata = { + .plat_start =NULL , + .init_quirk =NULL , + .suspend_quirk =NULL , + .resume_quirk =NULL , + .quirks =XHCI_SG_TRB_CACHE_SIZE_QUIRK , +};static int dwc3_host_get_irq (struct dwc3 *dwc) {
@@ -87,6 +96,11 @@ int dwc3_host_init(struct dwc3 *dwc) goto err; } + ret =platform_device_add_data (xhci, &dwc3_pdata,sizeof (dwc3_pdata)); +if (ret) { +dev_err (dwc->dev,"couldn't add platform data to xHCI device\n" ); + goto err; + }memset (props,0 ,sizeof (struct property_entry ) *ARRAY_SIZE (props));if (dwc->usb3_lpm_capable)
Signature · 1 line
2.28.0
Re: [PATCH v4 3/3] usb: dwc3: Pass quirk as platform data
Hello!
Quoted 4 lines · Tejas Joglekar
This commit adds the platform device data to setup the
XHCI_SG_TRB_CACHE_SIZE_QUIRKquirk. DWC3 hosts which are PCI devices does not use OF to create platform device but create xhci-plat platform device runtime. So
^ at
Quoted 1 line
this patch allow parent device to supply the quirk
Allows.
Quoted 12 lines
through platform data.
Signed-off-by Tejas Joglekar <[email protected]>--- drivers/usb/dwc3/host.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+)drivers/usb/dwc3/host.c +0
-0
drivers/usb/dwc3/host.c line 20
Quoted 10 lines
drivers/usb/dwc3/host.c +7
-0@@ -11,6 +11,15 @@#include <linux/platform_device.h> #include "core.h" +#include "../host/xhci-plat.h" + +static const struct xhci_plat_priv dwc3_pdata = { + .plat_start =NULL , + .init_quirk =NULL , + .suspend_quirk =NULL , + .resume_quirk =NULL ,
Why not rely on the compiler to fill these with zeros?
drivers/usb/dwc3/host.c line 25
Quoted 5 lines
drivers/usb/dwc3/host.c +2
-0+ .quirks = XHCI_SG_TRB_CACHE_SIZE_QUIRK, +};static int dwc3_host_get_irq (struct dwc3 *dwc) {
[...]
MBR, Sergei
Re: [PATCH v4 3/3] usb: dwc3: Pass quirk as platform data
Hi,
Quoted 8 lines · Sergei Shtylyov
Hello!
Quoted 4 lines · Tejas Joglekar
This commit adds the platform device data to setup the
XHCI_SG_TRB_CACHE_SIZE_QUIRKquirk. DWC3 hosts which are PCI devices does not use OF to create platform device but create xhci-plat platform device runtime. So
Quoted 6 lines
^ at
Quoted 1 line
this patch allow parent device to supply the quirk
Allows.
Noted.
Quoted 7 lines
Quoted 7 lines
through platform data.
Signed-off-by Tejas Joglekar <[email protected]>---
drivers/usb/dwc3/host.c| 14 ++++++++++++++1 file changed, 14 insertions(+)
Quoted 17 lines
Quoted 5 lines
drivers/usb/dwc3/host.c +0
-0drivers/usb/dwc3/host.c line 19
Quoted 9 lines
drivers/usb/dwc3/host.c +7
-0@@ -11,6 +11,15 @@ #include <linux/platform_device.h> #include "core.h" +#include "../host/xhci-plat.h" + +static const struct xhci_plat_priv dwc3_pdata = { + .plat_start = NULL, + .init_quirk = NULL, + .suspend_quirk = NULL, + .resume_quirk = NULL,Why not rely on the compiler to fill these with zeros?
Sure I will update
Quoted 7 lines
drivers/usb/dwc3/host.c line 25
Quoted 4 lines
drivers/usb/dwc3/host.c +2
-0+ .quirks = XHCI_SG_TRB_CACHE_SIZE_QUIRK, +}; static int dwc3_host_get_irq(struct dwc3 *dwc) {[...]
MBR, Sergei
Thanks & Regards,
Tejas Joglekar