usb: ss: Some fixes and cleanup for USB super-speed support
Subsystems (2)
v3 cover usb: ss: Some fixes and cleanup for USB super-speed support
Based on 'u-boot-usb' master branch.
This patch-series includes majorly some clean-up, few fixes and then some basic super-speed usb infrastructure addition, to help put support for XHCI in near future.
Changes from v2:
- Added a patch "usb: common: Weed out USB_**_PRINTFs from usb framework" to replace USB_***_PRINTF with debug() so as to get rid of unnecessary DEBUG macros in usb common framework and thereby rebasing other patches on top of this so that no USB_PRINTF appears further.
- Added a patch to reset only USB 2.0 ports, since USB 3.0 protocol ports don't require a reset to happen to move to 'enabled' state.
- Added a patch to move definition of 'min3()' out of ehci-hcd and putting the same as macro definition in common header.
- Using a 'switch-case' in portspeed() in cmd_usb.c
Changes from v1:
- Fixing the issue with 'ifno' as well as added 'if_desc'.
- Instead of turning-on only powered-off hub ports, power-cycle all the hub ports (aka: turning off and then turning on again power to all the ports.
- Fixing commenting style problem in
'usb: Update device class in usb device's descriptor'
- Fixing typo in commit message for
'usb: hub: Fix enumration timeout'
- Removing separate definition for 'struct usb_ep_desc'; thereby adding 'struct usb_ss_ep_comp_descriptor' to 'struct usb_interface' only. As a result modifying the patch accordingly also dropping the patch 'usb: eth: Fix for updated usb interface descriptor structure'
- Dropping the patch 'usb: hub: Increase device enumeration timeout for broken drives' for now (will come back with a solution at later point of time).
Vivek Gautam (8):
usb: common: Weed out USB_**_PRINTFs from usb framework
USB: Some cleanup prior to USB 3.0 interface addition
usb: hub: Power-cycle on root-hub ports
usb: Update device class in usb device's descriptor
usb: hub: Fix enumration timeout
USB: SS: Add support for Super Speed USB interface
usb: hub: Reset only usb 2.0 ports
usb: common: Use a global macro for 'min3'
common/cmd_usb.c | 24 +++- common/usb.c | 116 +++++++++--------- common/usb_hub.c | 228 ++++++++++++++++++++--------------- common/usb_kbd.c | 24 ++--- common/usb_storage.c | 283 +++++++++++++++++++++---------------------- drivers/usb/host/ehci-hcd.c | 10 -- include/common.h | 2 + include/usb.h | 6 + include/usb_defs.h | 26 ++++- 9 files changed, 389 insertions(+), 330 deletions(-)
Signature · 1 line
1.7.6.5
PATCH v3 1/8 usb: common: Weed out USB_**_PRINTFs from usb framework
USB_PRINTF, USB_HUB_PRINTF, USB_STOR_PRINTF, USB_KBD_PRINTF are nothing but conditional debug prints, depending on DEBUG. So better remove them and use debug() simply.
---
This patch added in V3(current-version) of this patch-series.
common/usb.c | 85 ++++++++---------- common/usb_hub.c | 162 +++++++++++++++----------------- common/usb_kbd.c | 24 ++--- common/usb_storage.c | 253 ++++++++++++++++++++++++-------------------------- 4 files changed, 241 insertions(+), 283 deletions(-)
common/usb.c +37 -48
@@ -57,17 +57,6 @@#include <asm/4xx_pci.h> #endif -#ifdef DEBUG -#define USB_DEBUG 1 -#define USB_HUB_DEBUG 1 -#else -#define USB_DEBUG 0 -#define USB_HUB_DEBUG 0 -#endif - -#define USB_PRINTF(fmt, args...) debug_cond(USB_DEBUG, fmt, ##args) -#define USB_HUB_PRINTF(fmt, args...) debug_cond(USB_HUB_DEBUG, fmt, ##args) -#define USB_BUFSIZ 512static struct usb_device usb_dev[USB_MAX_DEVICE ];
@@ -130,7 +119,7 @@ int usb_init(void) usb_started =1 ; } -USB_PRINTF ("scan end\n" ); +debug ("scan end\n" );/* if we were not able to find at least one working bus, bail out */ if (!usb_started) {puts ("USB error: all controllers failed lowlevel init\n" );
@@ -216,9 +205,9 @@ int usb_control_msg(struct usb_device *dev, unsigned int pipe, setup_packet->value =cpu_to_le16 (value); setup_packet->index =cpu_to_le16 (index); setup_packet->length =cpu_to_le16 (size); -USB_PRINTF ("usb_control_msg: request: 0x%X, requesttype: 0x%X, " \ -"value 0x%X index 0x%X length 0x%X\n" , - request, requesttype, value, index, size); +debug ("usb_control_msg: request: 0x%X, requesttype: 0x%X, " \ +"value 0x%X index 0x%X length 0x%X\n" , + request, requesttype, value, index, size); dev->status =USB_ST_NOT_PROC ;/*not yet processed */ if (submit_control_msg (dev, pipe, data, size, setup_packet) <0 )
@@ -314,22 +303,22 @@ usb_set_maxpacket_ep(struct usb_device *dev, int if_idx, int ep_idx)/* Control => bidirectional */ dev->epmaxpacketout[b] = ep_wMaxPacketSize; dev->epmaxpacketin[b] = ep_wMaxPacketSize; -USB_PRINTF ("##Control EP epmaxpacketout/in[%d] = %d\n" , - b, dev->epmaxpacketin[b]); +debug ("##Control EP epmaxpacketout/in[%d] = %d\n" , + b, dev->epmaxpacketin[b]); }else {if ((ep->bEndpointAddress &0x80 ) ==0 ) {/* OUT Endpoint */ if (ep_wMaxPacketSize > dev->epmaxpacketout[b]) { dev->epmaxpacketout[b] = ep_wMaxPacketSize; -USB_PRINTF ("##EP epmaxpacketout[%d] = %d\n" , - b, dev->epmaxpacketout[b]); +debug ("##EP epmaxpacketout[%d] = %d\n" , + b, dev->epmaxpacketout[b]); } }else {/* IN Endpoint */ if (ep_wMaxPacketSize > dev->epmaxpacketin[b]) { dev->epmaxpacketin[b] = ep_wMaxPacketSize; -USB_PRINTF ("##EP epmaxpacketin[%d] = %d\n" , - b, dev->epmaxpacketin[b]); +debug ("##EP epmaxpacketin[%d] = %d\n" , + b, dev->epmaxpacketin[b]); } }/* if out */ }/* if control */
@@ -358,7 +347,6 @@ static int usb_parse_config(struct usb_device *dev, {struct usb_descriptor_header *head;int index, ifno, epno, curr_if_num; -int i;u16 ep_wMaxPacketSize; ifno =-1 ;
@@ -414,23 +402,25 @@ static int usb_parse_config(struct usb_device *dev, if_desc[ifno].\ ep_desc[epno].\ wMaxPacketSize); -USB_PRINTF ("if %d, ep %d\n" , ifno, epno); +debug ("if %d, ep %d\n" , ifno, epno);break ;default :if (head->bLength ==0 )return 1 ; -USB_PRINTF ("unknown Description Type : %x\n" , - head->bDescriptorType); +debug ("unknown Description Type : %x\n" , + head->bDescriptorType); +#ifdef DEBUG { -#ifdef USB_DEBUG unsigned char *ch = (unsigned char *)head; -#endif +int i; +for (i =0 ; i < head->bLength; i++) -USB_PRINTF ("%02X " , *ch++); -USB_PRINTF ("\n\n\n" ); +debug ("%02X " , *ch++); +debug ("\n\n\n" ); } +#endif break ; } index += head->bLength;
@@ -514,8 +504,7 @@ int usb_get_configuration_no(struct usb_device *dev, } result =usb_get_descriptor (dev,USB_DT_CONFIG , cfgno, buffer, tmp); -USB_PRINTF ("get_conf_no %d Result %d, wLength %d\n" , - cfgno, result, tmp); +debug ("get_conf_no %d Result %d, wLength %d\n" , cfgno, result, tmp);return result; }
@@ -527,7 +516,7 @@ static int usb_set_address(struct usb_device *dev) {int res; -USB_PRINTF ("set address %d\n" , dev->devnum); +debug ("set address %d\n" , dev->devnum); res = usb_control_msg(dev, usb_snddefctrl(dev), USB_REQ_SET_ADDRESS, 0, (dev->devnum), 0,
@@ -579,7 +568,7 @@ int usb_set_interface(struct usb_device *dev, int interface, int alternate)static int usb_set_configuration (struct usb_device *dev,int configuration) {int res; -USB_PRINTF ("set configuration %d\n" , configuration); +debug ("set configuration %d\n" , configuration);/* set setup command */ res = usb_control_msg(dev, usb_sndctrlpipe(dev, 0), USB_REQ_SET_CONFIGURATION, 0,
@@ -731,19 +720,19 @@ int usb_string(struct usb_device *dev, int index, char *buf, size_t size)if (!dev->have_langid) { err =usb_string_sub (dev,0 ,0 , tbuf);if (err <0 ) { -USB_PRINTF ("error getting string descriptor 0 " \ -"(error=%lx)\n" , dev->status); +debug ("error getting string descriptor 0 " \ +"(error=%lx)\n" , dev->status);return -1 ; }else if (tbuf[0 ] <4 ) { -USB_PRINTF ("string descriptor 0 too short\n" ); +debug ("string descriptor 0 too short\n" );return -1 ; }else { dev->have_langid =-1 ; dev->string_langid = tbuf[2 ] | (tbuf[3 ] <<8 );/* always use the first langid listed */ -USB_PRINTF ("USB device number %d default " \ -"language ID 0x%x\n" , - dev->devnum, dev->string_langid); +debug ("USB device number %d default " \ +"language ID 0x%x\n" , + dev->devnum, dev->string_langid); } }
@@ -789,7 +778,7 @@ struct usb_device *usb_get_dev_index(int index)struct usb_device *usb_alloc_new_device (void *controller) {int i; -USB_PRINTF ("New Device %d\n" , dev_index); +debug ("New Device %d\n" , dev_index);if (dev_index ==USB_MAX_DEVICE ) {printf ("ERROR, too many USB Devices, max=%d\n" ,USB_MAX_DEVICE );return NULL ;
@@ -813,7 +802,7 @@ struct usb_device *usb_alloc_new_device(void *controller)void usb_free_device (void ) { dev_index--; -USB_PRINTF ("Freeing device node: %d\n" , dev_index); +debug ("Freeing device node: %d\n" , dev_index);memset (&usb_dev[dev_index],0 ,sizeof (struct usb_device )); usb_dev[dev_index].devnum =-1 ; }
@@ -880,7 +869,7 @@ int usb_new_device(struct usb_device *dev) err =usb_get_descriptor (dev,USB_DT_DEVICE ,0 , desc,64 );if (err <0 ) { -USB_PRINTF ("usb_new_device: usb_get_descriptor() failed\n" ); +debug ("usb_new_device: usb_get_descriptor() failed\n" );return 1 ; }
@@ -973,9 +962,9 @@ int usb_new_device(struct usb_device *dev) "len %d, status %lX\n", dev->act_len, dev->status);return -1 ; } -USB_PRINTF ("new device strings: Mfr=%d, Product=%d, SerialNumber=%d\n" , - dev->descriptor.iManufacturer, dev->descriptor.iProduct, - dev->descriptor.iSerialNumber); +debug ("new device strings: Mfr=%d, Product=%d, SerialNumber=%d\n" , + dev->descriptor.iManufacturer, dev->descriptor.iProduct, + dev->descriptor.iSerialNumber);memset (dev->mf,0 ,sizeof (dev->mf));memset (dev->prod,0 ,sizeof (dev->prod));memset (dev->serial,0 ,sizeof (dev->serial));
@@ -988,9 +977,9 @@ int usb_new_device(struct usb_device *dev)if (dev->descriptor.iSerialNumber)usb_string (dev, dev->descriptor.iSerialNumber, dev->serial,sizeof (dev->serial)); -USB_PRINTF ("Manufacturer %s\n" , dev->mf); -USB_PRINTF ("Product %s\n" , dev->prod); -USB_PRINTF ("SerialNumber %s\n" , dev->serial); +debug ("Manufacturer %s\n" , dev->mf); +debug ("Product %s\n" , dev->prod); +debug ("SerialNumber %s\n" , dev->serial);/* now prode if the device is a hub */ usb_hub_probe (dev,0 );return 0 ;
common/usb_hub.c +74 -88
331 lines not on this page · raw message
common/usb_kbd.c +9 -15
@@ -31,12 +31,6 @@#include <usb.h> -#ifdef USB_KBD_DEBUG -#define USB_KBD_PRINTF(fmt, args...) printf(fmt, ##args) -#else -#define USB_KBD_PRINTF(fmt, args...) -#endif -/* * If overwrite_console returns 1, the stdin, stderr and stdout * are switched to the serial port, else the settings in the
@@ -262,7 +256,7 @@ static int usb_kbd_translate(struct usb_kbd_pdata *data, unsigned char scancode,/* Report keycode if any */ if (keycode) { -USB_KBD_PRINTF ("%c" , keycode); +debug ("%c" , keycode);usb_kbd_put_queue (data, keycode); }
@@ -324,8 +318,8 @@ static int usb_kbd_irq_worker(struct usb_device *dev)static int usb_kbd_irq (struct usb_device *dev) {if ((dev->irq_status !=0 ) || (dev->irq_act_len !=8 )) { -USB_KBD_PRINTF ("USB KBD: Error %lX, len %d\n" , - dev->irq_status, dev->irq_act_len); +debug ("USB KBD: Error %lX, len %d\n" , + dev->irq_status, dev->irq_act_len);return 1 ; }
@@ -437,7 +431,7 @@ static int usb_kbd_probe(struct usb_device *dev, unsigned int ifnum)if ((ep->bmAttributes &3 ) !=3 )return 0 ; -USB_KBD_PRINTF ("USB KBD: found set protocol...\n" ); +debug ("USB KBD: found set protocol...\n" ); data =malloc (sizeof (struct usb_kbd_pdata ));if (!data) {
@@ -463,10 +457,10 @@ static int usb_kbd_probe(struct usb_device *dev, unsigned int ifnum)/* We found a USB Keyboard, install it. */ usb_set_protocol (dev, iface->desc.bInterfaceNumber,0 ); -USB_KBD_PRINTF ("USB KBD: found set idle...\n" ); +debug ("USB KBD: found set idle...\n" );usb_set_idle (dev, iface->desc.bInterfaceNumber,REPEAT_RATE ,0 ); -USB_KBD_PRINTF ("USB KBD: enable interrupt pipe...\n" ); +debug ("USB KBD: enable interrupt pipe...\n" );usb_submit_int_msg (dev, pipe, data->new, maxp >8 ?8 : maxp, ep->bInterval);
@@ -497,16 +491,16 @@ int drv_usb_kbd_init(void)continue ;/* We found a keyboard, check if it is already registered. */ -USB_KBD_PRINTF ("USB KBD: found set up device.\n" ); +debug ("USB KBD: found set up device.\n" ); old_dev =stdio_get_by_name (DEVNAME );if (old_dev) {/* Already registered, just return ok. */ -USB_KBD_PRINTF ("USB KBD: is already registered.\n" ); +debug ("USB KBD: is already registered.\n" );return 1 ; }/* Register the keyboard */ -USB_KBD_PRINTF ("USB KBD: register.\n" ); +debug ("USB KBD: register.\n" );memset (&usb_kbd_dev,0 ,sizeof (struct stdio_dev ));strcpy (usb_kbd_dev.name,DEVNAME ); usb_kbd_dev.flags =DEV_FLAGS_INPUT |DEV_FLAGS_SYSTEM ;
common/usb_storage.c +121 -132
645 lines not on this page · raw message
Signature · 1 line
1.7.6.5
PATCH v3 2/8 USB: Some cleanup prior to USB 3.0 interface addition
In reply to Vivek GautamSome cleanup in usb framework, nothing much on feature side.
---
Changes from v2:
- none
common/usb.c | 21 +++++++++++++-------- common/usb_storage.c | 30 ++++++++++++++++-------------- include/usb_defs.h | 2 +- 3 files changed, 30 insertions(+), 23 deletions(-)
common/usb.c +13 -8
@@ -348,6 +348,7 @@ static int usb_parse_config(struct usb_device *dev,struct usb_descriptor_header *head;int index, ifno, epno, curr_if_num;u16 ep_wMaxPacketSize; +struct usb_interface *if_desc =NULL ; ifno =-1 ; epno =-1 ;
@@ -375,23 +376,27 @@ static int usb_parse_config(struct usb_device *dev, &buffer[index])->bInterfaceNumber != curr_if_num) {/* this is a new interface, copy new desc */ ifno = dev->config.no_of_if; + if_desc = &dev->config.if_desc[ifno]; dev->config.no_of_if++; -memcpy (&dev->config.if_desc[ifno], - &buffer[index], buffer[index]); - dev->config.if_desc[ifno].no_of_ep =0 ; - dev->config.if_desc[ifno].num_altsetting =1 ; +memcpy (if_desc, &buffer[index], buffer[index]); + if_desc->no_of_ep =0 ; + if_desc->num_altsetting =1 ; curr_if_num = - dev->config.if_desc[ifno].desc.bInterfaceNumber; + if_desc->desc.bInterfaceNumber; } else {/* found alternate setting for the interface */ - dev->config.if_desc[ifno].num_altsetting++; +if (ifno >=0 ) { + if_desc = &dev->config.if_desc[ifno]; + if_desc->num_altsetting++; + } }break ;case USB_DT_ENDPOINT : epno = dev->config.if_desc[ifno].no_of_ep; + if_desc = &dev->config.if_desc[ifno];/* found an endpoint */ - dev->config.if_desc[ifno].no_of_ep++; -memcpy (&dev->config.if_desc[ifno].ep_desc[epno], + if_desc->no_of_ep++; +memcpy (&if_desc->ep_desc[epno], &buffer[index], buffer[index]); ep_wMaxPacketSize =get_unaligned (&dev->config.\ if_desc[ifno].\
common/usb_storage.c +16 -14
@@ -269,9 +269,9 @@ int usb_stor_scan(int mode) lun++) { usb_dev_desc[usb_max_devs].lun = lun;if (usb_stor_get_info (dev, &usb_stor[start], - &usb_dev_desc[usb_max_devs]) ==1 ) { - usb_max_devs++; - } + &usb_dev_desc[usb_max_devs]) ==1 ) { + usb_max_devs++; + } } }/* if storage device */
@@ -504,7 +504,7 @@ static int usb_stor_BBB_comdat(ccb *srb, struct us_data *us) dir_in =US_DIRECTION (srb->cmd[0 ]);#ifdef BBB_COMDAT_TRACE -printf ("dir %d lun %d cmdlen %d cmd %p datalen %d pdata %p\n" , +printf ("dir %d lun %d cmdlen %d cmd %p datalen %lu pdata %p\n" , dir_in, srb->lun, srb->cmdlen, srb->cmd, srb->datalen, srb->pdata);if (srb->cmdlen) {
@@ -1209,6 +1209,7 @@ int usb_storage_probe(struct usb_device *dev, unsigned int ifnum, {struct usb_interface *iface;int i; +struct usb_endpoint_descriptor *ep_desc;unsigned int flags =0 ;int protocol =0 ;
@@ -1291,24 +1292,25 @@ int usb_storage_probe(struct usb_device *dev, unsigned int ifnum,* We will ignore any others. */ for (i =0 ; i < iface->desc.bNumEndpoints; i++) { + ep_desc = &iface->ep_desc[i];/* is it an BULK endpoint? */ -if ((iface->ep_desc[i].bmAttributes & +if ((ep_desc->bmAttributes &USB_ENDPOINT_XFERTYPE_MASK ) ==USB_ENDPOINT_XFER_BULK ) { -if (iface->ep_desc[i].bEndpointAddress &USB_DIR_IN ) - ss->ep_in = iface->ep_desc[i].bEndpointAddress & -USB_ENDPOINT_NUMBER_MASK ; +if (ep_desc->bEndpointAddress &USB_DIR_IN ) + ss->ep_in = ep_desc->bEndpointAddress & +USB_ENDPOINT_NUMBER_MASK ;else ss->ep_out = - iface->ep_desc[i].bEndpointAddress & + ep_desc->bEndpointAddress &USB_ENDPOINT_NUMBER_MASK ; }/* is it an interrupt endpoint? */ -if ((iface->ep_desc[i].bmAttributes & -USB_ENDPOINT_XFERTYPE_MASK ) ==USB_ENDPOINT_XFER_INT ) { - ss->ep_int = iface->ep_desc[i].bEndpointAddress & -USB_ENDPOINT_NUMBER_MASK ; - ss->irqinterval = iface->ep_desc[i].bInterval; +if ((ep_desc->bmAttributes & +USB_ENDPOINT_XFERTYPE_MASK ) ==USB_ENDPOINT_XFER_INT ) { + ss->ep_int = ep_desc->bEndpointAddress & +USB_ENDPOINT_NUMBER_MASK ; + ss->irqinterval = ep_desc->bInterval; } } debug("Endpoints In %d Out %d Int %d\n",
include/usb_defs.h +1 -1
@@ -234,7 +234,7 @@#define HUB_CHAR_OCPM 0x0018/* -*Hub Status & Hub Change bit masks +* Hub Status & Hub Change bit masks */ #define HUB_STATUS_LOCAL_POWER 0x0001#define HUB_STATUS_OVERCURRENT 0x0002
Signature · 1 line
1.7.6.5
PATCH v3 3/8 usb: hub: Power-cycle on root-hub ports
In reply to Vivek GautamXHCI ports are powered on after a H/W reset, however EHCI ports are not. So disabling and re-enabling power on all ports invariably.
---
Changes from v2:
- Replaced USB_HUB_PRINTFs to debug()
common/usb_hub.c | 34 ++++++++++++++++++++++++++++++++++ 1 files changed, 34 insertions(+), 0 deletions(-)
common/usb_hub.c +34 -0
@@ -100,11 +100,45 @@ static void usb_hub_power_on(struct usb_hub_device *hub)int i;struct usb_device *dev;unsigned pgood_delay = hub->desc.bPwrOn2PwrGood *2 ; + ALLOC_CACHE_ALIGN_BUFFER(struct usb_port_status, portsts, 1); +unsigned short portstatus; +int ret; dev = hub->pusb_dev;/* Enable power to the ports */ debug ("enabling power on all ports\n" );for (i =0 ; i < dev->maxchild; i++) { +/* +* Power-cycle the ports here: aka, +* turning them off and turning on again. +*/ +usb_clear_port_feature (dev, i +1 ,USB_PORT_FEAT_POWER ); +debug ("port %d returns %lX\n" , i +1 , dev->status); + +/* Wait at least 2*bPwrOn2PwrGood for PP to change */ +mdelay (pgood_delay); + + ret =usb_get_port_status (dev, i +1 , portsts); +if (ret <0 ) { +debug ("port %d: get_port_status failed\n" , i +1 ); +return ; + } + +/* +* Check to confirm the state of Port Power: +* xHCI says "After modifying PP, s/w shall read +* PP and confirm that it has reached the desired state +* before modifying it again, undefined behavior may occur +* if this procedure is not followed". +* EHCI doesn't say anything like this, but no harm in keeping +* this. +*/ + portstatus =le16_to_cpu (portsts->wPortStatus); +if (portstatus & (USB_PORT_STAT_POWER <<1 )) { +debug ("port %d: Port power change failed\n" , i +1 ); +return ; + } +usb_set_port_feature (dev, i +1 ,USB_PORT_FEAT_POWER );debug ("port %d returns %lX\n" , i +1 , dev->status); }
Signature · 1 line
1.7.6.5
PATCH v3 3/8 usb: hub: Power-cycle on root-hub ports
Quoted 29 lines
XHCI ports are powered on after a H/W reset, however EHCI ports are not. So disabling and re-enabling power on all ports invariably.
Signed-off-by Amar <[email protected]>Signed-off-by Vivek Gautam <[email protected]>---
Changes from v2:
- ReplacedUSB_HUB_PRINTFstodebug()common/usb_hub.c | 34 ++++++++++++++++++++++++++++++++++ 1 files changed, 34 insertions(+), 0 deletions(-)common/usb_hub.c +3
-0@@ -100,11 +100,45 @@ static void usb_hub_power_on(struct usb_hub_device *hub)int i;struct usb_device *dev;unsigned pgood_delay = hub->desc.bPwrOn2PwrGood *2 ; + ALLOC_CACHE_ALIGN_BUFFER(struct usb_port_status, portsts, 1); +unsigned short portstatus; +int ret; dev = hub->pusb_dev;/* Enable power to the ports */ debug ("enabling power on all ports\n" );
common/usb_hub.c static void usb_hub_power_on(struct usb_hub_device *hub) · line 119
Quoted 10 lines
common/usb_hub.c +9
-0for (i =0 ; i < dev->maxchild; i++) { +/* +* Power-cycle the ports here: aka, +* turning them off and turning on again. +*/ +usb_clear_port_feature (dev, i +1 ,USB_PORT_FEAT_POWER ); +debug ("port %d returns %lX\n" , i +1 , dev->status); + +/* Wait at least 2*bPwrOn2PwrGood for PP to change */ +mdelay (pgood_delay);
This adds 20ms per port to USB boot times... is it really necessary? Why do we need to powercycle XHCI ports? Couldn't we just check if the ports are powered and only power them on if they aren't?
If you insist, please at least parallelize this. Disable power on all ports, wait, then check and reenable it.
common/usb_hub.c static void usb_hub_power_on(struct usb_hub_device *hub) · line 123
Quoted 4 lines
common/usb_hub.c +4
-0+ + ret =usb_get_port_status (dev, i +1 , portsts); +if (ret <0 ) { +debug ("port %d: get_port_status failed\n" , i +1 );
This is a severe error (because the ports won't come up again), so I think it should be a printf() instead of a debug().
Quoted 5 lines
common/usb_hub.c +5
-0+return ; + } + +/* +* Check to confirm the state of Port Power:
common/usb_hub.c static void usb_hub_power_on(struct usb_hub_device *hub) · line 138
Quoted 10 lines
common/usb_hub.c +10
-0+* xHCI says "After modifying PP, s/w shall read +* PP and confirm that it has reached the desired state +* before modifying it again, undefined behavior may occur +* if this procedure is not followed". +* EHCI doesn't say anything like this, but no harm in keeping +* this. +*/ + portstatus =le16_to_cpu (portsts->wPortStatus); +if (portstatus & (USB_PORT_STAT_POWER <<1 )) { +debug ("port %d: Port power change failed\n" , i +1 );
Same as above.
Quoted 6 lines
common/usb_hub.c +3
-0+return ; + } +usb_set_port_feature (dev, i +1 ,USB_PORT_FEAT_POWER );debug ("port %d returns %lX\n" , i +1 , dev->status); }
PATCH v3 3/8 usb: hub: Power-cycle on root-hub ports
HI Julius,
Quoted 35 lines · Julius Werner
Quoted 35 lines
XHCI ports are powered on after a H/W reset, however EHCI ports are not. So disabling and re-enabling power on all ports invariably.
Signed-off-by Amar <[email protected]>Signed-off-by Vivek Gautam <[email protected]>---
Changes from v2:
- ReplacedUSB_HUB_PRINTFstodebug()common/usb_hub.c | 34 ++++++++++++++++++++++++++++++++++ 1 files changed, 34 insertions(+), 0 deletions(-)common/usb_hub.c +8
-0@@ -100,11 +100,45 @@ static void usb_hub_power_on(struct usb_hub_device *hub)int i;struct usb_device *dev;unsigned pgood_delay = hub->desc.bPwrOn2PwrGood *2 ; + ALLOC_CACHE_ALIGN_BUFFER(struct usb_port_status, portsts, 1); +unsigned short portstatus; +int ret; dev = hub->pusb_dev;/* Enable power to the ports */ debug ("enabling power on all ports\n" );for (i =0 ; i < dev->maxchild; i++) { +/* +* Power-cycle the ports here: aka, +* turning them off and turning on again. +*/ +usb_clear_port_feature (dev, i +1 ,USB_PORT_FEAT_POWER );
Quoted 8 lines
common/usb_hub.c static void usb_hub_power_on(struct usb_hub_device *hub) · line 119
Quoted 4 lines
common/usb_hub.c +4
-0+debug ("port %d returns %lX\n" , i +1 , dev->status); + +/* Wait at least 2*bPwrOn2PwrGood for PP to change */ +mdelay (pgood_delay);This adds 20ms per port to USB boot times... is it really necessary? Why do we need to powercycle XHCI ports? Couldn't we just check if the ports are powered and only power them on if they aren't?
This 20ms delay is truely being taken to be on safer side (twice of Power-on-to-power-good),
its not observational.
In my earlier patch-series, we were doing the same way as you are suggesting here (power-on
ports only if they aren't), however Marek suggested to power-cycle the ports. This would ensure
that we don't have any spurious Port status (telling us that port is powered up).
Actually i was seeing a strage behavior on USB 2.0 protocol ports available with XHCI.
Since all ports with XHCI are powered-up after a Chip-reset, the instant we do a power-on
on these ports (as with original code - simply setting the PORT_POWER feature), the Connect status
change bit used to get cleared (however Current connect status bit was still set).
So we arrived at solution that either we don't power-up XHCI ports or do a power-cycle on them.
Quoted 3 lines
If you insist, please at least parallelize this. Disable power on all ports, wait, then check and reenable it.
Hmm, so right now we are doing this for one port at a time.
I am sure parallelising things as you suggested would be best to do here, but can you please explain, would handling one port at a time lead to unwanted behavior from Host's side somehow ?
Quoted 8 lines
common/usb_hub.c static void usb_hub_power_on(struct usb_hub_device *hub) · line 123
Quoted 5 lines
common/usb_hub.c +4
-0+ + ret =usb_get_port_status (dev, i +1 , portsts); +if (ret <0 ) { +debug ("port %d: get_port_status failed\n" , i +1 );This is a severe error (because the ports won't come up again), so I think it should be a
printf()instead of adebug().
Sure, will change this to pritnf()
Quoted 17 lines
Quoted 6 lines
common/usb_hub.c +5
-0+return ; + } + +/* +* Check to confirm the state of Port Power: common/usb_hub.c static void usb_hub_power_on(struct usb_hub_device *hub) · line 138
Quoted 10 lines
common/usb_hub.c +10
-0+* xHCI says "After modifying PP, s/w shall read +* PP and confirm that it has reached the desired state +* before modifying it again, undefined behavior may occur +* if this procedure is not followed". +* EHCI doesn't say anything like this, but no harm in keeping +* this. +*/ + portstatus =le16_to_cpu (portsts->wPortStatus); +if (portstatus & (USB_PORT_STAT_POWER <<1 )) { +debug ("port %d: Port power change failed\n" , i +1 );
Quoted 1 line
Same as above.
Sure, will change this.
Quoted 7 lines
Quoted 7 lines
common/usb_hub.c +3
-0+return ; + } +usb_set_port_feature (dev, i +1 ,USB_PORT_FEAT_POWER );debug ("port %d returns %lX\n" , i +1 , dev->status); }
Signature · 2 lines
Best Regards Vivek
PATCH v3 3/8 usb: hub: Power-cycle on root-hub ports
Quoted 9 lines
This 20ms delay is truely being taken to be on safer side (twice of Power-on-to-power-good),
its not observational.
In my earlier patch-series, we were doing the same way as you are suggesting here (power-on
ports only if they aren't), however Marek suggested to power-cycle the ports. This would ensure
that we don't have any spurious Port status (telling us that port is powered up).
Fair enough... I guess 20ms overall is not a big deal in the whole picture. I sometimes tend to overoptimize things...
Quoted 8 lines
Actually i was seeing a strage behavior on USB 2.0 protocol ports available with XHCI.
Since all ports with XHCI are powered-up after a Chip-reset, the instant we do a power-on
on these ports (as with original code - simply setting thePORT_POWERfeature), the Connect status
change bit used to get cleared (however Current connect status bit was still set).
This is a bug in your XHCI code I hadn't spotted before: in xhci_submit_root(SET_FEATURE) you read the PORTSC register, add a bit, and write it again... without calling xhci_port_state_to_neutral() inbetween. Thus you clear any pending change events when you set PORT_POWER. (Seems the EHCI code has a similar bug in CLEAR_FEATURE, now that I'm looking at it... someone should put a 'reg &= ~EHCI_PS_CLEAR;' in there.)
Quoted 4 lines
Hmm, so right now we are doing this for one port at a time.
I am sure parallelising things as you suggested would be best to do here, but can you please explain, would handling one port at a time lead to unwanted behavior from Host's side somehow ?
It doesn't affect correctness, just the amount of time "wasted". Doing it one port at a time means you wait 100ms on a 5 port root hub, while you could get by with 20ms overall by parallelizing it.
PATCH v3 3/8 usb: hub: Power-cycle on root-hub ports
Hi,
Quoted 22 lines · Julius Werner
Quoted 9 lines
This 20ms delay is truely being taken to be on safer side (twice of Power-on-to-power-good),
its not observational.
In my earlier patch-series, we were doing the same way as you are suggesting here (power-on
ports only if they aren't), however Marek suggested to power-cycle the ports. This would ensure
that we don't have any spurious Port status (telling us that port is powered up).Fair enough... I guess 20ms overall is not a big deal in the whole picture. I sometimes tend to overoptimize things...
Quoted 8 lines
Actually i was seeing a strage behavior on USB 2.0 protocol ports available with XHCI.
Since all ports with XHCI are powered-up after a Chip-reset, the instant we do a power-on
on these ports (as with original code - simply setting thePORT_POWERfeature), the Connect status
change bit used to get cleared (however Current connect status bit was still set).
Quoted 5 lines
This is a bug in your XHCI code I hadn't spotted before: in
xhci_submit_root(SET_FEATURE)you read the PORTSC register, add a bit, and write it again... without callingxhci_port_state_to_neutral()inbetween. Thus you clear any pending change events when you setPORT_POWER.
Right, we need to do that.
Quoted 3 lines
(Seems the EHCI code has a similar bug in
CLEAR_FEATURE, now that I'm looking at it... someone should put a 'reg &= ~EHCI_PS_CLEAR;' in there.)
True, EHCI has it for SET_FEATURE but not for CLEAR_FEATURE.
Quoted 9 lines
Quoted 5 lines
Hmm, so right now we are doing this for one port at a time.
I am sure parallelising things as you suggested would be best to do here, but can you please explain, would handling one port at a time lead to unwanted behavior from Host's side somehow ?It doesn't affect correctness, just the amount of time "wasted". Doing it one port at a time means you wait 100ms on a 5 port root hub, while you could get by with 20ms overall by parallelizing it.
True, will amend this as required.
Signature · 2 lines
Best Regards Vivek
PATCH v3 4/8 usb: Update device class in usb device's descriptor
In reply to Vivek GautamFetch the device class into usb device's dwcriptors, so that the host controller's driver can use this info to differentiate between HUB and DEVICE.
---
Changes from v2:
- none
common/usb.c | 5 +++++ 1 files changed, 5 insertions(+), 0 deletions(-)
common/usb.c +5 -0
@@ -879,6 +879,11 @@ int usb_new_device(struct usb_device *dev) } dev->descriptor.bMaxPacketSize0 = desc->bMaxPacketSize0; +/* +* Fetch the device class, driver can use this info +* to differentiate between HUB and DEVICE. +*/ + dev->descriptor.bDeviceClass = desc->bDeviceClass;/* find the port number we're at */ if (parent) {
Signature · 1 line
1.7.6.5
PATCH v3 4/8 usb: Update device class in usb device's descriptor
Quoted 18 lines
Fetch the device class into usb device's dwcriptors, so that the host controller's driver can use this info to differentiate between HUB and DEVICE.
Signed-off-by Amar <[email protected]>---
Changes from v2:
- nonecommon/usb.c | 5 +++++ 1 files changed, 5 insertions(+), 0 deletions(-)common/usb.c +0
-0
common/usb.c int usb_new_device(struct usb_device *dev) · line 886
Quoted 8 lines
common/usb.c +5
-0@@ -879,6 +879,11 @@ int usb_new_device(struct usb_device *dev) } dev->descriptor.bMaxPacketSize0 = desc->bMaxPacketSize0; +/* +* Fetch the device class, driver can use this info +* to differentiate between HUB and DEVICE. +*/ + dev->descriptor.bDeviceClass = desc->bDeviceClass;
There is nothing wrong with this, but I also don't think you need it. In those places where you check for the hub class in your XHCI stack, you actually want to check whether the device is the root hub... so you should do that differently to avoid breaking external hubs (e.g. by comparing the device address). I think we should drop this since there is no real use for it.
Quoted 3 lines
common/usb.c +0
-0/* find the port number we're at */ if (parent) {
PATCH v3 5/8 usb: hub: Fix enumration timeout
In reply to Vivek GautamPatch b6d7852c increases timeout for enumeration, taking worst case to be 10 sec.get_timer() api returns timestamp in milliseconds, which is what we are checking in the do-while() loop in usb_hub_configure() (get_timer(start) < CONFIG_SYS_HZ * 10).
This should give us a required check for 10 seconds, and thereby we don't need to add additional mdelay of 100 microseconds in each cycle.
---
Changes from v2:
- none
common/usb_hub.c | 1 - 1 files changed, 0 insertions(+), 1 deletions(-)
common/usb_hub.c +0 -1
@@ -443,7 +443,6 @@ static int usb_hub_configure(struct usb_device *dev) (portstatus & USB_PORT_STAT_CONNECTION))break ; -mdelay (100 ); }while (get_timer (start) <CONFIG_SYS_HZ *10 );if (ret <0 )
Signature · 1 line
1.7.6.5
PATCH v3 6/8 USB: SS: Add support for Super Speed USB interface
In reply to Vivek GautamThis adds usb framework support for super-speed usb, which will further facilitate to add stack support for xHCI.
---
Changes from v2:
- Replacing if-else with switch-case in portspeed() in cmd_usb.c
common/cmd_usb.c | 24 ++++++++++++++++++------ common/usb.c | 5 +++++ common/usb_hub.c | 8 ++++++-- include/usb.h | 6 ++++++ include/usb_defs.h | 24 +++++++++++++++++++++++- 5 files changed, 58 insertions(+), 9 deletions(-)
common/cmd_usb.c +18 -6
@@ -271,12 +271,24 @@ static void usb_display_config(struct usb_device *dev)static inline char *portspeed (int speed) { -if (speed ==USB_SPEED_HIGH ) -return "480 Mb/s" ; -else if (speed ==USB_SPEED_LOW ) -return "1.5 Mb/s" ; -else -return "12 Mb/s" ; +char *speed_str; + +switch (speed) { +case USB_SPEED_SUPER : + speed_str ="5 Gb/s" ; +break ; +case USB_SPEED_HIGH : + speed_str ="480 Mb/s" ; +break ; +case USB_SPEED_LOW : + speed_str ="1.5 Mb/s" ; +break ; +default : + speed_str ="12 Mb/s" ; +break ; + } + +return speed_str; }/* shows the device tree recursively */
common/usb.c +5 -0
@@ -409,6 +409,11 @@ static int usb_parse_config(struct usb_device *dev, wMaxPacketSize);debug ("if %d, ep %d\n" , ifno, epno);break ; +case USB_DT_SS_ENDPOINT_COMP : + if_desc = &dev->config.if_desc[ifno]; +memcpy (&if_desc->ss_ep_comp_desc[epno], + &buffer[index], buffer[index]); +break ;default :if (head->bLength ==0 )return 1 ;
common/usb_hub.c +6 -2
@@ -165,7 +165,9 @@ static struct usb_hub_device *usb_hub_allocate(void)static inline char *portspeed (int portstatus) { -if (portstatus & (1 <<USB_PORT_FEAT_HIGHSPEED )) +if (portstatus & (1 <<USB_PORT_FEAT_SUPERSPEED )) +return "5 Gb/s" ; +else if (portstatus & (1 <<USB_PORT_FEAT_HIGHSPEED ))return "480 Mb/s" ;else if (portstatus & (1 <<USB_PORT_FEAT_LOWSPEED ))return "1.5 Mb/s" ;
@@ -268,7 +270,9 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port)/* Allocate a new device struct for it */ usb =usb_alloc_new_device (dev->controller); -if (portstatus &USB_PORT_STAT_HIGH_SPEED ) +if (portstatus &USB_PORT_STAT_SUPER_SPEED ) + usb->speed =USB_SPEED_SUPER ; +else if (portstatus &USB_PORT_STAT_HIGH_SPEED ) usb->speed =USB_SPEED_HIGH ;else if (portstatus &USB_PORT_STAT_LOW_SPEED ) usb->speed =USB_SPEED_LOW ;
include/usb.h +6 -0
@@ -76,6 +76,12 @@ struct usb_interface {unsigned char act_altsetting;struct usb_endpoint_descriptor ep_desc[USB_MAXENDPOINTS ]; +/* +* Super Speed Device will have Super Speed Endpoint +* Companion Descriptor (section 9.6.7 of usb 3.0 spec) +* Revision 1.0 June 6th 2011 +*/ +struct usb_ss_ep_comp_descriptor ss_ep_comp_desc[USB_MAXENDPOINTS ]; } __attribute__ ((packed));/* Configuration information.. */
include/usb_defs.h +23 -1
@@ -203,6 +203,8 @@#define USB_PORT_FEAT_POWER 8#define USB_PORT_FEAT_LOWSPEED 9#define USB_PORT_FEAT_HIGHSPEED 10 +#define USB_PORT_FEAT_FULLSPEED 11 +#define USB_PORT_FEAT_SUPERSPEED 12#define USB_PORT_FEAT_C_CONNECTION 16#define USB_PORT_FEAT_C_ENABLE 17#define USB_PORT_FEAT_C_SUSPEND 18
@@ -218,8 +220,20 @@#define USB_PORT_STAT_POWER 0x0100#define USB_PORT_STAT_LOW_SPEED 0x0200#define USB_PORT_STAT_HIGH_SPEED 0x0400/* support for EHCI */ +#define USB_PORT_STAT_FULL_SPEED 0x0800 +#define USB_PORT_STAT_SUPER_SPEED 0x1000/* support for XHCI */ #define USB_PORT_STAT_SPEED \ - (USB_PORT_STAT_LOW_SPEED | USB_PORT_STAT_HIGH_SPEED) + (USB_PORT_STAT_LOW_SPEED | USB_PORT_STAT_HIGH_SPEED | \ + USB_PORT_STAT_FULL_SPEED | USB_PORT_STAT_SUPER_SPEED) + +/* +* Additions to wPortStatus bit field from USB 3.0 +* See USB 3.0 spec Table 10-10 +*/ +#define USB_PORT_STAT_LINK_STATE 0x01e0 +#define USB_SS_PORT_STAT_POWER 0x0200 +#define USB_SS_PORT_STAT_SPEED 0x1c00 +#define USB_PORT_STAT_SPEED_5GBPS 0x0000/* wPortChange bits */ #define USB_PORT_STAT_C_CONNECTION 0x0001
@@ -228,6 +242,14 @@#define USB_PORT_STAT_C_OVERCURRENT 0x0008#define USB_PORT_STAT_C_RESET 0x0010 +/* +* Addition to wPortChange bit fields form USB 3.0 +* See USB 3.0 spec Table 10-11 +*/ +#define USB_PORT_STAT_C_BH_RESET 0x0020 +#define USB_PORT_STAT_C_LINK_STATE 0x0040 +#define USB_PORT_STAT_C_CONFIG_ERROR 0x0080 +/* wHubCharacteristics (masks) */ #define HUB_CHAR_LPSM 0x0003#define HUB_CHAR_COMPOUND 0x0004
Signature · 1 line
1.7.6.5
PATCH v3 6/8 USB: SS: Add support for Super Speed USB interface
Migrating my comments here for public discussion.
Quoted 73 lines
This adds usb framework support for super-speed usb, which will further facilitate to add stack support for xHCI.
Signed-off-by Vikas C Sajjan <[email protected]>Signed-off-by Vivek Gautam <[email protected]>---
Changes from v2:
- Replacing if-else with switch-case inportspeed()incmd_usb.ccommon/cmd_usb.c | 24 ++++++++++++++++++------ common/usb.c | 5 +++++ common/usb_hub.c | 8 ++++++-- include/usb.h | 6 ++++++ include/usb_defs.h | 24 +++++++++++++++++++++++- 5 files changed, 58 insertions(+), 9 deletions(-)common/cmd_usb.c +18
-6@@ -271,12 +271,24 @@ static void usb_display_config(struct usb_device *dev)static inline char *portspeed (int speed) { -if (speed ==USB_SPEED_HIGH ) -return "480 Mb/s" ; -else if (speed ==USB_SPEED_LOW ) -return "1.5 Mb/s" ; -else -return "12 Mb/s" ; +char *speed_str; + +switch (speed) { +case USB_SPEED_SUPER : + speed_str ="5 Gb/s" ; +break ; +case USB_SPEED_HIGH : + speed_str ="480 Mb/s" ; +break ; +case USB_SPEED_LOW : + speed_str ="1.5 Mb/s" ; +break ; +default : + speed_str ="12 Mb/s" ; +break ; + } + +return speed_str; }/* shows the device tree recursively */ common/usb.c +5
-0@@ -409,6 +409,11 @@ static int usb_parse_config(struct usb_device *dev, wMaxPacketSize);debug ("if %d, ep %d\n" , ifno, epno);break ; +case USB_DT_SS_ENDPOINT_COMP : + if_desc = &dev->config.if_desc[ifno]; +memcpy (&if_desc->ss_ep_comp_desc[epno], + &buffer[index], buffer[index]); +break ;default :if (head->bLength ==0 )return 1 ;common/usb_hub.c +0
-0
common/usb_hub.c static struct usb_hub_device *usb_hub_allocate(void) · line 168
Quoted 5 lines
common/usb_hub.c +1
-1@@ -165,7 +165,9 @@ static struct usb_hub_device *usb_hub_allocate(void)static inline char *portspeed (int portstatus) { -if (portstatus & (1 <<USB_PORT_FEAT_HIGHSPEED )) +if (portstatus & (1 <<USB_PORT_FEAT_SUPERSPEED ))
It doesn't make sense to use the USB_PORT_FEAT_ constants to read a port status here. These should be USB_PORT_STAT_ instead (you could use a switch statement if you mask and shift them correctly).
Quoted 6 lines
common/usb_hub.c +2
-0+return "5 Gb/s" ; + else if (portstatus & (1 << USB_PORT_FEAT_HIGHSPEED))return "480 Mb/s" ; else if (portstatus & (1 << USB_PORT_FEAT_LOWSPEED))return "1.5 Mb/s" ;
common/usb_hub.c void usb_hub_port_connect_change(struct usb_device *dev, int port) · line 278
Quoted 10 lines
common/usb_hub.c +3
-1@@ -268,7 +270,9 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port)/* Allocate a new device struct for it */ usb =usb_alloc_new_device (dev->controller); -if (portstatus &USB_PORT_STAT_HIGH_SPEED ) +if (portstatus &USB_PORT_STAT_SUPER_SPEED ) + usb->speed =USB_SPEED_SUPER ; +else if (portstatus &USB_PORT_STAT_HIGH_SPEED ) usb->speed =USB_SPEED_HIGH ;else if (portstatus &USB_PORT_STAT_LOW_SPEED ) usb->speed =USB_SPEED_LOW ;
Should change to switch statement after implementing my suggestion to usb_defs.h
Quoted 22 lines
include/usb.h +6
-0@@ -76,6 +76,12 @@ struct usb_interface {unsigned char act_altsetting;struct usb_endpoint_descriptor ep_desc[USB_MAXENDPOINTS ]; +/* +* Super Speed Device will have Super Speed Endpoint +* Companion Descriptor (section 9.6.7 of usb 3.0 spec) +* Revision 1.0 June 6th 2011 +*/ +struct usb_ss_ep_comp_descriptor ss_ep_comp_desc[USB_MAXENDPOINTS ]; } __attribute__ ((packed));/* Configuration information.. */ include/usb_defs.h +0
-0
include/usb_defs.h line 207
Quoted 5 lines
include/usb_defs.h +2
-0@@ -203,6 +203,8 @@#define USB_PORT_FEAT_POWER 8#define USB_PORT_FEAT_LOWSPEED 9#define USB_PORT_FEAT_HIGHSPEED 10 +#define USB_PORT_FEAT_FULLSPEED 11 +#define USB_PORT_FEAT_SUPERSPEED 12
Speed is never set as a feature anyway. We should just get rid of all of these and always use USB_PORT_STAT_ for them.
include/usb_defs.h line 224
Quoted 9 lines
include/usb_defs.h +2
-0#define USB_PORT_FEAT_C_CONNECTION 16#define USB_PORT_FEAT_C_ENABLE 17#define USB_PORT_FEAT_C_SUSPEND 18@@ -218,8 +220,20 @@#define USB_PORT_STAT_POWER 0x0100#define USB_PORT_STAT_LOW_SPEED 0x0200#define USB_PORT_STAT_HIGH_SPEED 0x0400/* support for EHCI */ +#define USB_PORT_STAT_FULL_SPEED 0x0800 +#define USB_PORT_STAT_SUPER_SPEED 0x1000/* support for XHCI */
These two new values are *not* correct. 0x800 is actually PORT_TEST, and 0x1000 is PORT_INDICATOR. Full speed has always been reported by having both the LOW_SPEED and the HIGH_SPEED bits zero, and you should define it that way (for use in switch statements).
If you want to fake SuperSpeed into the USB 2.0 hub descriptor (see my comment on xhci_port_speed in the unreleased CL), you could represent it with 0xc00, which should be a reserved value for all existing 1.0 and 2.0 hubs.
include/usb_defs.h line 227
Quoted 4 lines
include/usb_defs.h +2
-1#define USB_PORT_STAT_SPEED \ - (USB_PORT_STAT_LOW_SPEED | USB_PORT_STAT_HIGH_SPEED) + (USB_PORT_STAT_LOW_SPEED | USB_PORT_STAT_HIGH_SPEED | \ + USB_PORT_STAT_FULL_SPEED | USB_PORT_STAT_SUPER_SPEED)
Maybe this should be renamed to USB_PORT_STAT_SPEED_MASK to make its purpose clearer.
include/usb_defs.h line 230
Quoted 3 lines
include/usb_defs.h +3
-0+ +/* +* Additions to wPortStatus bit field from USB 3.0
I would reword "Additions" to "Changes" and explicitly state that these fields concern the new USB 3.0 hub descriptor format, which is different and exclusively applies to SuperSpeed hubs.
include/usb_defs.h line 231
Quoted 1 line
include/usb_defs.h +1
-0+* See USB 3.0 spec Table 10-10
In my spec it's Table 10-11... are you sure you have the most recent version?
include/usb_defs.h line 233
Quoted 2 lines
include/usb_defs.h +2
-0+*/ +#define USB_PORT_STAT_LINK_STATE 0x01e0
For consistency, I would prefix all values in here with USB_SS_PORT_STAT_ (to make extra clear that they belong to something different than the ones above).
Quoted 7 lines
include/usb_defs.h +3
-0+#define USB_SS_PORT_STAT_POWER 0x0200 +#define USB_SS_PORT_STAT_SPEED 0x1c00 +#define USB_PORT_STAT_SPEED_5GBPS 0x0000/* wPortChange bits */ #define USB_PORT_STAT_C_CONNECTION 0x0001
include/usb_defs.h line 249
Quoted 8 lines
include/usb_defs.h +5
-0@@ -228,6 +242,14 @@#define USB_PORT_STAT_C_OVERCURRENT 0x0008#define USB_PORT_STAT_C_RESET 0x0010 +/* +* Addition to wPortChange bit fields form USB 3.0 +* See USB 3.0 spec Table 10-11 +*/ +#define USB_PORT_STAT_C_BH_RESET 0x0020
Same here. New incompatible port status format for SuperSpeed hubs, please make the comment clearer and prefix as USB_SS_PORT_STAT_C_
Quoted 6 lines
include/usb_defs.h +3
-0+#define USB_PORT_STAT_C_LINK_STATE 0x0040 +#define USB_PORT_STAT_C_CONFIG_ERROR 0x0080 +/* wHubCharacteristics (masks) */ #define HUB_CHAR_LPSM 0x0003#define HUB_CHAR_COMPOUND 0x0004
PATCH v3 6/8 USB: SS: Add support for Super Speed USB interface
Hi Julius,
Quoted 1 line
Migrating my comments here for public discussion.
Maybe you can make a patch(set) against u-boot-usb/next ?
Best regards,
Marek Vasut
PATCH v3 6/8 USB: SS: Add support for Super Speed USB interface
These patches haven't gone in yet, right? I think Vivek wants to discuss/update them himself, he just asked me to move my reviews to this thread.
PATCH v3 6/8 USB: SS: Add support for Super Speed USB interface
Dear Julius Werner,
Quoted 3 lines
These patches haven't gone in yet, right? I think Vivek wants to discuss/update them himself, he just asked me to move my reviews to this thread.
They're not in, but they're already pushed in my tree. It'd be also easier to review diff instead of full patches again.
Best regards,
Marek Vasut
PATCH v3 6/8 USB: SS: Add support for Super Speed USB interface
Hi Marek,
Quoted 8 lines · Marek Vasut
Dear Julius Werner,
Quoted 3 lines
These patches haven't gone in yet, right? I think Vivek wants to discuss/update them himself, he just asked me to move my reviews to this thread.
They're not in, but they're already pushed in my tree. It'd be also easier to review diff instead of full patches again.
I shall send a diff patchset for these changes, but do we have a possibility of squashing the changes
to earler commits at some point of time later, so that clean changes get to u-boot ?
Or we shall go as you suggest.
Signature · 2 lines
Best Regards Vivek
PATCH v3 6/8 USB: SS: Add support for Super Speed USB interface
Dear Vivek Gautam,
Quoted 9 lines
Hi Marek,
Quoted 6 lines · Marek Vasut
Dear Julius Werner,
Quoted 3 lines
These patches haven't gone in yet, right? I think Vivek wants to discuss/update them himself, he just asked me to move my reviews to this thread.
Quoted 8 lines
Quoted 2 lines
They're not in, but they're already pushed in my tree. It'd be also easier to review diff instead of full patches again.
I shall send a diff patchset for these changes, but do we have a possibility of squashing the changes
to earler commits at some point of time later, so that clean changes get to u-boot ?
Or we shall go as you suggest.
Just send a subsequent patch ;-)
Best regards,
Marek Vasut
PATCH v3 6/8 USB: SS: Add support for Super Speed USB interface
Hi Marek,
Quoted 14 lines · Marek Vasut
Dear Vivek Gautam,
Quoted 12 lines
Hi Marek,
Quoted 8 lines · Marek Vasut
Dear Julius Werner,
Quoted 3 lines
These patches haven't gone in yet, right? I think Vivek wants to discuss/update them himself, he just asked me to move my reviews to this thread.
They're not in, but they're already pushed in my tree. It'd be also easier to review diff instead of full patches again.
Quoted 7 lines
Quoted 5 lines
I shall send a diff patchset for these changes, but do we have a possibility of squashing the changes
to earler commits at some point of time later, so that clean changes get to u-boot ?
Or we shall go as you suggest.Just send a subsequent patch ;-)
Sure, will send subsequent patches on u-boot-usb/next
Quoted 3 lines
Best regards,
Marek Vasut
Signature · 2 lines
Best Regards Vivek
PATCH v3 6/8 USB: SS: Add support for Super Speed USB interface
In reply to Julius WernerHi Julius,
Quoted 1 line · Julius Werner
Migrating my comments here for public discussion.
Thanks for your valuable comments here.
Quoted 75 lines
Quoted 75 lines
This adds usb framework support for super-speed usb, which will further facilitate to add stack support for xHCI.
Signed-off-by Vikas C Sajjan <[email protected]>Signed-off-by Vivek Gautam <[email protected]>---
Changes from v2:
- Replacing if-else with switch-case inportspeed()incmd_usb.ccommon/cmd_usb.c | 24 ++++++++++++++++++------ common/usb.c | 5 +++++ common/usb_hub.c | 8 ++++++-- include/usb.h | 6 ++++++ include/usb_defs.h | 24 +++++++++++++++++++++++- 5 files changed, 58 insertions(+), 9 deletions(-)common/cmd_usb.c +18
-6@@ -271,12 +271,24 @@ static void usb_display_config(struct usb_device *dev)static inline char *portspeed (int speed) { -if (speed ==USB_SPEED_HIGH ) -return "480 Mb/s" ; -else if (speed ==USB_SPEED_LOW ) -return "1.5 Mb/s" ; -else -return "12 Mb/s" ; +char *speed_str; + +switch (speed) { +case USB_SPEED_SUPER : + speed_str ="5 Gb/s" ; +break ; +case USB_SPEED_HIGH : + speed_str ="480 Mb/s" ; +break ; +case USB_SPEED_LOW : + speed_str ="1.5 Mb/s" ; +break ; +default : + speed_str ="12 Mb/s" ; +break ; + } + +return speed_str; }/* shows the device tree recursively */ common/usb.c +5
-0@@ -409,6 +409,11 @@ static int usb_parse_config(struct usb_device *dev, wMaxPacketSize);debug ("if %d, ep %d\n" , ifno, epno);break ; +case USB_DT_SS_ENDPOINT_COMP : + if_desc = &dev->config.if_desc[ifno]; +memcpy (&if_desc->ss_ep_comp_desc[epno], + &buffer[index], buffer[index]); +break ;default :if (head->bLength ==0 )return 1 ;common/usb_hub.c +0
-0@@ -165,7 +165,9 @@ static struct usb_hub_device *usb_hub_allocate(void)
Quoted 8 lines
common/usb_hub.c static struct usb_hub_device *usb_hub_allocate(void) · line 168
Quoted 4 lines
common/usb_hub.c +1
-1static inline char *portspeed (int portstatus) { -if (portstatus & (1 <<USB_PORT_FEAT_HIGHSPEED )) +if (portstatus & (1 <<USB_PORT_FEAT_SUPERSPEED ))It doesn't make sense to use the
USB_PORT_FEAT_constants to read a port status here. These should beUSB_PORT_STAT_instead (you could use a switch statement if you mask and shift them correctly).
True, we should be 'ANDing' here with respectve PORT_STAT_ constants, will modify this as required.
Quoted 10 lines
Quoted 10 lines
common/usb_hub.c +2
-0+return "5 Gb/s" ; + else if (portstatus & (1 << USB_PORT_FEAT_HIGHSPEED))return "480 Mb/s" ; else if (portstatus & (1 << USB_PORT_FEAT_LOWSPEED))return "1.5 Mb/s" ;@@ -268,7 +270,9 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port)/* Allocate a new device struct for it */ usb =usb_alloc_new_device (dev->controller);
Quoted 10 lines
common/usb_hub.c void usb_hub_port_connect_change(struct usb_device *dev, int port) · line 278
Quoted 7 lines
common/usb_hub.c +3
-1-if (portstatus &USB_PORT_STAT_HIGH_SPEED ) +if (portstatus &USB_PORT_STAT_SUPER_SPEED ) + usb->speed =USB_SPEED_SUPER ; +else if (portstatus &USB_PORT_STAT_HIGH_SPEED ) usb->speed =USB_SPEED_HIGH ;else if (portstatus &USB_PORT_STAT_LOW_SPEED ) usb->speed =USB_SPEED_LOW ;Should change to switch statement after implementing my suggestion to
usb_defs.h
Sure,
Quoted 17 lines
Quoted 17 lines
include/usb.h +6
-0@@ -76,6 +76,12 @@ struct usb_interface {unsigned char act_altsetting;struct usb_endpoint_descriptor ep_desc[USB_MAXENDPOINTS ]; +/* +* Super Speed Device will have Super Speed Endpoint +* Companion Descriptor (section 9.6.7 of usb 3.0 spec) +* Revision 1.0 June 6th 2011 +*/ +struct usb_ss_ep_comp_descriptor ss_ep_comp_desc[USB_MAXENDPOINTS ]; } __attribute__ ((packed));
Quoted 14 lines
include/usb_defs.h line 207
Quoted 11 lines
include/usb.h +0
-0/* Configuration information.. */ include/usb_defs.h +2
-0@@ -203,6 +203,8 @@#define USB_PORT_FEAT_POWER 8#define USB_PORT_FEAT_LOWSPEED 9#define USB_PORT_FEAT_HIGHSPEED 10 +#define USB_PORT_FEAT_FULLSPEED 11 +#define USB_PORT_FEAT_SUPERSPEED 12Speed is never set as a feature anyway. We should just get rid of all of these and always use
USB_PORT_STAT_for them.
These are simply 'Hub class feature numbers' given in USB 2.0 and 3.0 specs. Why don't we just keep them like that (I shall be removing the USB_PORT_FEAT_FULLSPEED
and USB_PORT_FEAT_SUPERSPEED definitely from here). In case we want to check the port status field,
we would check against PORT_STAT_ constants.
Quoted 15 lines
include/usb_defs.h line 224
Quoted 10 lines
include/usb_defs.h +2
-0#define USB_PORT_FEAT_C_CONNECTION 16#define USB_PORT_FEAT_C_ENABLE 17#define USB_PORT_FEAT_C_SUSPEND 18@@ -218,8 +220,20 @@#define USB_PORT_STAT_POWER 0x0100#define USB_PORT_STAT_LOW_SPEED 0x0200#define USB_PORT_STAT_HIGH_SPEED 0x0400/* support for EHCI */ +#define USB_PORT_STAT_FULL_SPEED 0x0800 +#define USB_PORT_STAT_SUPER_SPEED 0x1000/* support for XHCI */ These two new values are *not* correct.
0x800is actuallyPORT_TEST, and0x1000isPORT_INDICATOR. Full speed has always been reported by having both theLOW_SPEEDand theHIGH_SPEEDbits zero, and you should define it that way (for use in switch statements).
Sure, will remove these two constants.
Quoted 5 lines
If you want to fake SuperSpeed into the USB 2.0 hub descriptor (see my comment on
xhci_port_speedin the unreleased CL), you could represent it with0xc00, which should be a reserved value for all existing 1.0 and 2.0 hubs.
Quoted 7 lines
include/usb_defs.h line 227
Quoted 4 lines
include/usb_defs.h +2
-1#define USB_PORT_STAT_SPEED \ - (USB_PORT_STAT_LOW_SPEED | USB_PORT_STAT_HIGH_SPEED) + (USB_PORT_STAT_LOW_SPEED | USB_PORT_STAT_HIGH_SPEED | \ + USB_PORT_STAT_FULL_SPEED | USB_PORT_STAT_SUPER_SPEED)Maybe this should be renamed to
USB_PORT_STAT_SPEED_MASKto make its purpose clearer.
Alright, shall rename it to USB_PORT_STAT_SPEED_MASK.
Quoted 8 lines
include/usb_defs.h line 230
Quoted 4 lines
include/usb_defs.h +3
-0+ +/* +* Additions to wPortStatus bit field from USB 3.0 I would reword "Additions" to "Changes" and explicitly state that these fields concern the new USB 3.0 hub descriptor format, which is different and exclusively applies to SuperSpeed hubs.
This bit fields are actually additions to USB 3.0 framework. These fields are not defined in
USB 2.0 specs, or else they are reserved there. If you insist i can modify this as suggested.
Quoted 4 lines
include/usb_defs.h line 231
Quoted 2 lines
include/usb_defs.h +1
-0+* See USB 3.0 spec Table 10-10 In my spec it's Table 10-11... are you sure you have the most recent version?
Yea, my mistake, typo !!
Quoted 6 lines
include/usb_defs.h line 233
Quoted 3 lines
include/usb_defs.h +2
-0+*/ +#define USB_PORT_STAT_LINK_STATE 0x01e0For consistency, I would prefix all values in here with
USB_SS_PORT_STAT_(to make extra clear that they belong to something different than the ones above).
True, shall change this to make things consistent.
Quoted 11 lines
Quoted 11 lines
include/usb_defs.h +3
-0+#define USB_SS_PORT_STAT_POWER 0x0200 +#define USB_SS_PORT_STAT_SPEED 0x1c00 +#define USB_PORT_STAT_SPEED_5GBPS 0x0000/* wPortChange bits */ #define USB_PORT_STAT_C_CONNECTION 0x0001@@ -228,6 +242,14 @@#define USB_PORT_STAT_C_OVERCURRENT 0x0008#define USB_PORT_STAT_C_RESET 0x0010
Quoted 8 lines
include/usb_defs.h line 249
Quoted 5 lines
include/usb_defs.h +5
-0+/* +* Addition to wPortChange bit fields form USB 3.0 +* See USB 3.0 spec Table 10-11 +*/ +#define USB_PORT_STAT_C_BH_RESET 0x0020Same here. New incompatible port status format for SuperSpeed hubs, please make the comment clearer and prefix as
USB_SS_PORT_STAT_C_
Sure, will add the relavant prefix, and write proper comment.
Quoted 7 lines
Quoted 7 lines
include/usb_defs.h +3
-0+#define USB_PORT_STAT_C_LINK_STATE 0x0040 +#define USB_PORT_STAT_C_CONFIG_ERROR 0x0080 +/* wHubCharacteristics (masks) */ #define HUB_CHAR_LPSM 0x0003#define HUB_CHAR_COMPOUND 0x0004
Signature · 2 lines
Best Regards Vivek
PATCH v3 6/8 USB: SS: Add support for Super Speed USB interface
Quoted 8 lines
Quoted 3 lines
I would reword "Additions" to "Changes" and explicitly state that these fields concern the new USB 3.0 hub descriptor format, which is different and exclusively applies to SuperSpeed hubs.
This bit fields are actually additions to USB 3.0 framework. These fields are not defined in
USB 2.0 specs, or else they are reserved there. If you insist i can modify this as suggested.
Yes, okay, I think we mean the same thing here. I just wanted to point out that the USB 3.0 port status is actually a completely different thing from the USB 2.0 port status (which is still valid for non-SuperSpeed hubs in USB 3.0 environments, even though the spec hasn't printed it out again). They did not just *add* a few bits to the old status bit field, they actually *changed* existing bits (not just reserved ones)... e.g. bit 9 means LOW_SPEED for USB 2.0 hubs but PORT_POWER for SuperSpeed hubs. I found this very confusing when I first read it, so I thought it might be worth making it extra obvious through comments.
PATCH v3 6/8 USB: SS: Add support for Super Speed USB interface
Hi Julius,
Quoted 9 lines · Julius Werner
Quoted 8 lines
Quoted 3 lines
I would reword "Additions" to "Changes" and explicitly state that these fields concern the new USB 3.0 hub descriptor format, which is different and exclusively applies to SuperSpeed hubs.
This bit fields are actually additions to USB 3.0 framework. These fields are not defined in
USB 2.0 specs, or else they are reserved there. If you insist i can modify this as suggested.
Quoted 10 lines
Yes, okay, I think we mean the same thing here. I just wanted to point out that the USB 3.0 port status is actually a completely different thing from the USB 2.0 port status (which is still valid for non-SuperSpeed hubs in USB 3.0 environments, even though the spec hasn't printed it out again). They did not just *add* a few bits to the old status bit field, they actually *changed* existing bits (not just reserved ones)... e.g. bit 9 means
LOW_SPEEDfor USB 2.0 hubs butPORT_POWERfor SuperSpeed hubs. I found this very confusing when I first read it, so I thought it might be worth making it extra obvious through comments.
Fair enough, i shall change this accordingly, Thanks !!
Signature · 2 lines
Best Regards Vivek
PATCH v3 7/8 usb: hub: Reset only usb 2.0 ports
In reply to Vivek GautamAs per XHCI specifications USB 3.0 protocol ports attempt to advance to 'Enabled' state; however USB 2.0 protocol ports require software reset to advance them to 'Enabled' state. Thereby, inferring that software need to reset USB 2.0 protocol ports invariably (as per EHCI spec or xHCI spec).
---
This patch added in V3(current-version) of this patch-series.
common/usb_hub.c | 23 +++++++++++++++++++---- 1 files changed, 19 insertions(+), 4 deletions(-)
common/usb_hub.c +19 -4
@@ -233,6 +233,7 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port)struct usb_device *usb; ALLOC_CACHE_ALIGN_BUFFER(struct usb_port_status, portsts, 1);unsigned short portstatus; +uint32_t do_port_reset =1 ;/* Check status */ if (usb_get_port_status (dev, port +1 , portsts) <0 ) {
@@ -246,6 +247,9 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port)le16_to_cpu (portsts->wPortChange), portspeed(portstatus)); +if (portstatus & (1 <<USB_PORT_FEAT_SUPERSPEED )) + do_port_reset =0 ; +/* Clear the connection change status */ usb_clear_port_feature (dev, port +1 ,USB_PORT_FEAT_C_CONNECTION );
@@ -259,10 +263,21 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port) }mdelay (200 ); -/* Reset the port */ -if (hub_port_reset (dev, port, &portstatus) <0 ) { -printf ("cannot reset port %i!?\n" , port +1 ); -return ; +/* +* Reset the port: +* As per xHCI protocol, USB 3.0 devices do not require +* a port reset, however USB 2.0 device do require the same +* to let ports proceed to 'enabled' state +* +* XXX: Will this break EHCI ?? +* probably not, above condition for 'do_port_reset' checks for +* speed, and for EHCI it can't reach Super speed anyways. +*/ +if (do_port_reset) { +if (hub_port_reset (dev, port, &portstatus) <0 ) { +printf ("cannot reset port %i!?\n" , port +1 ); +return ; + } }mdelay (200 );
Signature · 1 line
1.7.6.5
PATCH v3 7/8 usb: hub: Reset only usb 2.0 ports
Sorry, forgot this one yesterday. I would consider to just drop/revert this patch entirely. It's not wrong, but it adds complexity where it is not needed. You don't have to reset SuperSpeed devices, but it shouldn't hurt either and from what I can tell Linux does it as well.
Quoted 62 lines
As per XHCI specifications USB 3.0 protocol ports attempt to advance to 'Enabled' state; however USB 2.0 protocol ports require software reset to advance them to 'Enabled' state. Thereby, inferring that software need to reset USB 2.0 protocol ports invariably (as per EHCI spec or xHCI spec).
Signed-off-by Vivek Gautam <[email protected]>---
This patch added in V3(current-version) of this patch-series.common/usb_hub.c | 23 +++++++++++++++++++---- 1 files changed, 19 insertions(+), 4 deletions(-)common/usb_hub.c +19
-4@@ -233,6 +233,7 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port)struct usb_device *usb; ALLOC_CACHE_ALIGN_BUFFER(struct usb_port_status, portsts, 1);unsigned short portstatus; +uint32_t do_port_reset =1 ;/* Check status */ if (usb_get_port_status (dev, port +1 , portsts) <0 ) {@@ -246,6 +247,9 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port)le16_to_cpu (portsts->wPortChange), portspeed(portstatus)); +if (portstatus & (1 <<USB_PORT_FEAT_SUPERSPEED )) + do_port_reset =0 ; +/* Clear the connection change status */ usb_clear_port_feature (dev, port +1 ,USB_PORT_FEAT_C_CONNECTION );@@ -259,10 +263,21 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port) }mdelay (200 ); -/* Reset the port */ -if (hub_port_reset (dev, port, &portstatus) <0 ) { -printf ("cannot reset port %i!?\n" , port +1 ); -return ; +/* +* Reset the port: +* As per xHCI protocol, USB 3.0 devices do not require +* a port reset, however USB 2.0 device do require the same +* to let ports proceed to 'enabled' state +* +* XXX: Will this break EHCI ?? +* probably not, above condition for 'do_port_reset' checks for +* speed, and for EHCI it can't reach Super speed anyways. +*/ +if (do_port_reset) { +if (hub_port_reset (dev, port, &portstatus) <0 ) { +printf ("cannot reset port %i!?\n" , port +1 ); +return ; + } }mdelay (200 );
PATCH v3 7/8 usb: hub: Reset only usb 2.0 ports
Dear Julius Werner,
Quoted 4 lines
Sorry, forgot this one yesterday. I would consider to just drop/revert this patch entirely. It's not wrong, but it adds complexity where it is not needed. You don't have to reset SuperSpeed devices, but it shouldn't hurt either and from what I can tell Linux does it as well.
Ok, I can drop this one.
Quoted 69 lines
Quoted 69 lines
As per XHCI specifications USB 3.0 protocol ports attempt to advance to 'Enabled' state; however USB 2.0 protocol ports require software reset to advance them to 'Enabled' state. Thereby, inferring that software need to reset USB 2.0 protocol ports invariably (as per EHCI spec or xHCI spec).
Signed-off-by Vivek Gautam <[email protected]>---
This patch added in V3(current-version) of this patch-series.common/usb_hub.c | 23 +++++++++++++++++++---- 1 files changed, 19 insertions(+), 4 deletions(-)common/usb_hub.c +1
-0@@ -233,6 +233,7 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port)struct usb_device *usb; ALLOC_CACHE_ALIGN_BUFFER(struct usb_port_status, portsts, 1);unsigned short portstatus; +uint32_t do_port_reset =1 ;common/usb_hub.c +3
-0/* Check status */ if (usb_get_port_status (dev, port +1 , portsts) <0 ) {@@ -246,6 +247,9 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port)le16_to_cpu (portsts->wPortChange), portspeed(portstatus)); +if (portstatus & (1 <<USB_PORT_FEAT_SUPERSPEED )) + do_port_reset =0 ; +common/usb_hub.c +15
-4/* Clear the connection change status */ usb_clear_port_feature (dev, port +1 ,USB_PORT_FEAT_C_CONNECTION );@@ -259,10 +263,21 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port) }mdelay (200 ); -/* Reset the port */ -if (hub_port_reset (dev, port, &portstatus) <0 ) { -printf ("cannot reset port %i!?\n" , port +1 ); -return ; +/* +* Reset the port: +* As per xHCI protocol, USB 3.0 devices do not require +* a port reset, however USB 2.0 device do require the same +* to let ports proceed to 'enabled' state +* +* XXX: Will this break EHCI ?? +* probably not, above condition for 'do_port_reset' checks for +* speed, and for EHCI it can't reach Super speed anyways. +*/ +if (do_port_reset) { +if (hub_port_reset (dev, port, &portstatus) <0 ) { +printf ("cannot reset port %i!?\n" , port +1 ); +return ; + }
Quoted 3 lines
common/usb_hub.c void usb_hub_port_connect_change(struct usb_device *dev, int port) · line 283
Quoted 3 lines
common/usb_hub.c +0
-0}mdelay (200 );
Best regards,
Marek Vasut
PATCH v3 7/8 usb: hub: Reset only usb 2.0 ports
Quoted 8 lines · Marek Vasut
Dear Julius Werner,
Quoted 4 lines
Sorry, forgot this one yesterday. I would consider to just drop/revert this patch entirely. It's not wrong, but it adds complexity where it is not needed. You don't have to reset SuperSpeed devices, but it shouldn't hurt either and from what I can tell Linux does it as well.
Ok, I can drop this one.
Alright, we will drop this.
I think a possible bug related to calling xhci_port_state_to_neutral() in xhci_submit_root()
(one pointed by Julius in [PATCH v3 3/8] usb: hub: Power-cycle on root-hub ports) was causing
USB 3.0 protocol ports fail while resetting.
Now things are fine.
Quoted 80 lines
Quoted 70 lines
Quoted 70 lines
As per XHCI specifications USB 3.0 protocol ports attempt to advance to 'Enabled' state; however USB 2.0 protocol ports require software reset to advance them to 'Enabled' state. Thereby, inferring that software need to reset USB 2.0 protocol ports invariably (as per EHCI spec or xHCI spec).
Signed-off-by Vivek Gautam <[email protected]>---
This patch added in V3(current-version) of this patch-series.common/usb_hub.c | 23 +++++++++++++++++++---- 1 files changed, 19 insertions(+), 4 deletions(-)common/usb_hub.c +1
-0@@ -233,6 +233,7 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port)struct usb_device *usb; ALLOC_CACHE_ALIGN_BUFFER(struct usb_port_status, portsts, 1);unsigned short portstatus; +uint32_t do_port_reset =1 ;/* Check status */ if (usb_get_port_status(dev, port + 1, portsts) < 0) {common/usb_hub.c +3
-0@@ -246,6 +247,9 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port)le16_to_cpu (portsts->wPortChange), portspeed(portstatus)); +if (portstatus & (1 <<USB_PORT_FEAT_SUPERSPEED )) + do_port_reset =0 ; +/* Clear the connection change status */ usb_clear_port_feature(dev, port + 1, USB_PORT_FEAT_C_CONNECTION);common/usb_hub.c +15
-4@@ -259,10 +263,21 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port) }mdelay (200 ); -/* Reset the port */ -if (hub_port_reset (dev, port, &portstatus) <0 ) { -printf ("cannot reset port %i!?\n" , port +1 ); -return ; +/* +* Reset the port: +* As per xHCI protocol, USB 3.0 devices do not require +* a port reset, however USB 2.0 device do require the same +* to let ports proceed to 'enabled' state +* +* XXX: Will this break EHCI ?? +* probably not, above condition for 'do_port_reset' checks for +* speed, and for EHCI it can't reach Super speed anyways. +*/ +if (do_port_reset) { +if (hub_port_reset (dev, port, &portstatus) <0 ) { +printf ("cannot reset port %i!?\n" , port +1 ); +return ; + }Quoted 3 lines
Quoted 3 lines
} mdelay(200);Best regards,
Marek Vasut
_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
http://lists.denx.de/mailman/listinfo/u-boot
Signature · 2 lines
Best Regards Vivek
PATCH v3 7/8 usb: hub: Reset only usb 2.0 ports
Hi Marek,
Quoted 9 lines · Vivek Gautam
Quoted 8 lines · Marek Vasut
Dear Julius Werner,
Quoted 4 lines
Sorry, forgot this one yesterday. I would consider to just drop/revert this patch entirely. It's not wrong, but it adds complexity where it is not needed. You don't have to reset SuperSpeed devices, but it shouldn't hurt either and from what I can tell Linux does it as well.
Ok, I can drop this one.
We need to drop this from u-boot-usb/next branch, right ? I am sure this is added to you to-do list, just wanted to make sure since i could
see it in 'next' right now.
Thanks !!
Quoted 82 lines
Alright, we will drop this.
I think a possible bug related to callingxhci_port_state_to_neutral()inxhci_submit_root()
(one pointed by Julius in [PATCH v3 3/8] usb: hub: Power-cycle on root-hub ports) was causing
USB 3.0 protocol ports fail while resetting.
Now things are fine.Quoted 73 lines
Quoted 73 lines
Quoted 73 lines
As per XHCI specifications USB 3.0 protocol ports attempt to advance to 'Enabled' state; however USB 2.0 protocol ports require software reset to advance them to 'Enabled' state. Thereby, inferring that software need to reset USB 2.0 protocol ports invariably (as per EHCI spec or xHCI spec).
Signed-off-by Vivek Gautam <[email protected]>---
This patch added in V3(current-version) of this patch-series.common/usb_hub.c | 23 +++++++++++++++++++---- 1 files changed, 19 insertions(+), 4 deletions(-)common/usb_hub.c +1
-0@@ -233,6 +233,7 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port)struct usb_device *usb; ALLOC_CACHE_ALIGN_BUFFER(struct usb_port_status, portsts, 1);unsigned short portstatus; +uint32_t do_port_reset =1 ;/* Check status */ if (usb_get_port_status(dev, port + 1, portsts) < 0) {common/usb_hub.c +3
-0@@ -246,6 +247,9 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port)le16_to_cpu (portsts->wPortChange), portspeed(portstatus)); +if (portstatus & (1 <<USB_PORT_FEAT_SUPERSPEED )) + do_port_reset =0 ; +/* Clear the connection change status */ usb_clear_port_feature(dev, port + 1, USB_PORT_FEAT_C_CONNECTION);common/usb_hub.c +15
-4@@ -259,10 +263,21 @@ void usb_hub_port_connect_change(struct usb_device *dev, int port) }mdelay (200 ); -/* Reset the port */ -if (hub_port_reset (dev, port, &portstatus) <0 ) { -printf ("cannot reset port %i!?\n" , port +1 ); -return ; +/* +* Reset the port: +* As per xHCI protocol, USB 3.0 devices do not require +* a port reset, however USB 2.0 device do require the same +* to let ports proceed to 'enabled' state +* +* XXX: Will this break EHCI ?? +* probably not, above condition for 'do_port_reset' checks for +* speed, and for EHCI it can't reach Super speed anyways. +*/ +if (do_port_reset) { +if (hub_port_reset (dev, port, &portstatus) <0 ) { +printf ("cannot reset port %i!?\n" , port +1 ); +return ; + }} mdelay(200);
Signature · 2 lines
Best Regards Vivek
PATCH v3 7/8 usb: hub: Reset only usb 2.0 ports
Dear Vivek Gautam,
Quoted 14 lines
Hi Marek,
On Wed, Apr 24, 2013 at 11:38 AM, Vivek Gautam
Quoted 8 lines · <[email protected]> wrote:
Quoted 7 lines · Marek Vasut
Dear Julius Werner,
Quoted 4 lines
Sorry, forgot this one yesterday. I would consider to just drop/revert this patch entirely. It's not wrong, but it adds complexity where it is not needed. You don't have to reset SuperSpeed devices, but it shouldn't hurt either and from what I can tell Linux does it as well.
Quoted 7 lines
Quoted 1 line
Quoted 1 line
Ok, I can drop this one.
We need to drop this from u-boot-usb/next branch, right ? I am sure this is added to you to-do list, just wanted to make sure since i could
see it in 'next' right now.
Thanks !!
Dropped and pushed to -next .
Best regards,
Marek Vasut
PATCH v3 8/8 usb: common: Use a global macro for 'min3'
In reply to Vivek GautamWe can use a common global macro for calculating minimum of 3 numbers. Put the same in 'common header' and let 'ehci' use it.
---
This patch added in V3(current-version) of this patch-series.
drivers/usb/host/ehci-hcd.c | 10 ---------- include/common.h | 2 ++ 2 files changed, 2 insertions(+), 10 deletions(-)
drivers/usb/host/ehci-hcd.c +0 -10
@@ -587,16 +587,6 @@ fail:return -1 ; } -static inline int min3 (int a,int b,int c) -{ - -if (b < a) - a = b; -if (c < a) - a = c; -return a; -} -int ehci_submit_root (struct usb_device *dev,unsigned long pipe,void *buffer,int length,struct devrequest *req)
include/common.h +2 -0
@@ -211,6 +211,8 @@ typedef void (interrupt_handler_t)(void *);#define MIN (x, y) min(x, y)#define MAX (x, y) max(x, y) +#define min3 (a, b, c) min(min(a, b), c) +/* * Return the absolute value of a number. *
Signature · 1 line
1.7.6.5
PATCH v3 8/8 usb: common: Use a global macro for 'min3'
Dear Vivek Gautam,
Quoted 39 lines
We can use a common global macro for calculating minimum of 3 numbers. Put the same in 'common header' and let 'ehci' use it.
Signed-off-by Vivek Gautam <[email protected]>---
This patch added in V3(current-version) of this patch-series.
drivers/usb/host/ehci-hcd.c | 10 ---------- include/common.h | 2 ++ 2 files changed, 2 insertions(+), 10 deletions(-)drivers/usb/host/ehci-hcd.c +0
-10@@ -587,16 +587,6 @@ fail:return -1 ; } -static inline int min3 (int a,int b,int c) -{ - -if (b < a) - a = b; -if (c < a) - a = c; -return a; -} -int ehci_submit_root (struct usb_device *dev,unsigned long pipe,void *buffer,int length,struct devrequest *req)include/common.h +0
-0
include/common.h typedef void (interrupt_handler_t)(void *); · line 215
Quoted 5 lines
include/common.h +2
-0@@ -211,6 +211,8 @@ typedef void (interrupt_handler_t)(void *);#define MIN (x, y) min(x, y)#define MAX (x, y) max(x, y) +#define min3 (a, b, c) min(min(a, b), c) +
You might want to keep it as an inline-function to allow GCC do the type- checking?
include/common.h typedef void (interrupt_handler_t)(void *); · line 218
Quoted 3 lines
include/common.h +0
-0/* * Return the absolute value of a number. *
Best regards,
Marek Vasut
PATCH 1/1 usb: common: Use a global definition for 'min3'
We can use a common global method for calculating minimum of 3 numbers. Put the same in 'common header' and let 'ehci' use it.
--- drivers/usb/host/ehci-hcd.c | 10 ---------- include/common.h | 5 +++++ 2 files changed, 5 insertions(+), 10 deletions(-)
drivers/usb/host/ehci-hcd.c +0 -10
@@ -603,16 +603,6 @@ fail:return -1 ; } -static inline int min3 (int a,int b,int c) -{ - -if (b < a) - a = b; -if (c < a) - a = c; -return a; -} -int ehci_submit_root (struct usb_device *dev,unsigned long pipe,void *buffer,int length,struct devrequest *req)
include/common.h +5 -0
@@ -211,6 +211,11 @@ typedef void (interrupt_handler_t)(void *);#define MIN (x, y) min(x, y)#define MAX (x, y) max(x, y) +static inline int min3 (int a,int b,int c) +{ +return min (min (a, b), c); +} +/* * Return the absolute value of a number. *
Signature · 1 line
1.7.6.5
Marek Vasut
AppliedPATCH 1/1 usb: common: Use a global definition for 'min3'
Dear Vivek Gautam,
Quoted 5 lines
We can use a common global method for calculating minimum of 3 numbers. Put the same in 'common header' and let 'ehci' use it.
Signed-off-by Vivek Gautam <[email protected]>
Applied, thanks
Quoted 32 lines
--- drivers/usb/host/ehci-hcd.c | 10 ---------- include/common.h | 5 +++++ 2 files changed, 5 insertions(+), 10 deletions(-)drivers/usb/host/ehci-hcd.c +0
-10@@ -603,16 +603,6 @@ fail:return -1 ; } -static inline int min3 (int a,int b,int c) -{ - -if (b < a) - a = b; -if (c < a) - a = c; -return a; -} -int ehci_submit_root (struct usb_device *dev,unsigned long pipe,void *buffer,int length,struct devrequest *req)include/common.h +0
-0@@ -211,6 +211,11 @@ typedef void (interrupt_handler_t)(void *);#define MIN (x, y) min(x, y)
include/common.h typedef void (interrupt_handler_t)(void *); · line 221
Quoted 10 lines
include/common.h +5
-0#define MAX (x, y) max(x, y) +static inline int min3 (int a,int b,int c) +{ +return min (min (a, b), c); +} +/* * Return the absolute value of a number. *
Best regards,
Marek Vasut
Tom Rini
NAKPATCH 1/1 usb: common: Use a global definition for 'min3'
Quoted 9 lines · Marek Vasut
Dear Vivek Gautam,
Quoted 5 lines
We can use a common global method for calculating minimum of 3 numbers. Put the same in 'common header' and let 'ehci' use it.
Signed-off-by Vivek Gautam <[email protected]>Applied, thanks
NAK, sorry. Lets re-sync with the kernel's min/max/min3/max3 defines here instead.
Signature · 8 lines
Tom -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 836 bytes Desc: Digital signature URL: <http://lists.denx.de/pipermail/u-boot/attachments/20130422/8c39b4fe/attachment.pgp>
PATCH 1/1 usb: common: Use a global definition for 'min3'
Quoted 7 lines · Tom Rini
Quoted 6 lines · Marek Vasut
Dear Vivek Gautam,
Quoted 4 lines
We can use a common global method for calculating minimum of 3 numbers. Put the same in 'common header' and let 'ehci' use it.
Quoted 6 lines
Quoted 3 lines
Quoted 1 line
Signed-off-by Vivek Gautam <[email protected]>Applied, thanks
NAK, sorry. Lets re-sync with the kernel's min/max/min3/max3 defines here instead.
Alright, i shall sync the definitions with Linux kernel and send the patch for same.
Signature · 2 lines
Best Regards Vivek
PATCH 1/1 usb: common: Use a global definition for 'min3'
Hi Marek,
Quoted 10 lines · Vivek Gautam
Quoted 9 lines · Tom Rini
Quoted 8 lines · Marek Vasut
Dear Vivek Gautam,
Quoted 5 lines
We can use a common global method for calculating minimum of 3 numbers. Put the same in 'common header' and let 'ehci' use it.
Signed-off-by Vivek Gautam <[email protected]>
Quoted 7 lines
Quoted 4 lines
Quoted 1 line
Applied, thanks
NAK, sorry. Lets re-sync with the kernel's min/max/min3/max3 defines here instead.
Alright, i shall sync the definitions with Linux kernel and send the patch for same.
Will you be dropping this patch from u-boot-usb/next or shall i send a subsequent patch on
top of this ?
Signature · 2 lines
Best Regards Vivek
PATCH 1/1 usb: common: Use a global definition for 'min3'
Dear Vivek Gautam,
Quoted 21 lines
Hi Marek,
On Wed, Apr 24, 2013 at 11:49 AM, Vivek Gautam
Quoted 15 lines · <[email protected]> wrote:
Quoted 13 lines · Tom Rini
Quoted 9 lines · Marek Vasut
Dear Vivek Gautam,
Quoted 5 lines
We can use a common global method for calculating minimum of 3 numbers. Put the same in 'common header' and let 'ehci' use it.
Signed-off-by Vivek Gautam <[email protected]>Applied, thanks
NAK, sorry. Lets re-sync with the kernel's min/max/min3/max3 defines here instead.
Quoted 6 lines
Quoted 2 lines
Alright, i shall sync the definitions with Linux kernel and send the patch for same.
Will you be dropping this patch from u-boot-usb/next or shall i send a subsequent patch on
top of this ?
Dropped and pushed
Best regards,
Marek Vasut
v3 cover usb: ss: Some fixes and cleanup for USB super-speed support
In reply to Vivek GautamDear Vivek Gautam,
Quoted 17 lines
Based on 'u-boot-usb' master branch.
This patch-series includes majorly some clean-up, few fixes and then some basic super-speed usb infrastructure addition, to help put support for XHCI in near future.
Changes from v2:
- Added a patch "usb: common: Weed out USB_**_PRINTFs from usb framework" to replace USB_***_PRINTF withdebug()so as to get rid of unnecessary DEBUG macros in usb common framework and thereby rebasing other patches on top of this so that noUSB_PRINTFappears further.
- Added a patch to reset only USB 2.0 ports, since USB 3.0 protocol ports don't require a reset to happen to move to 'enabled' state.
- Added a patch to move definition of 'min3()' out of ehci-hcd and putting the same as macro definition in common header.
- Using a 'switch-case' inportspeed()incmd_usb.c
Quoted 16 lines
Changes from v1:
- Fixing the issue with 'ifno' as well as added 'if_desc'.
- Instead of turning-on only powered-off hub ports, power-cycle all the hub ports (aka: turning off and then turning on again power to all the ports.
- Fixing commenting style problem in
'usb: Update device class in usb device's descriptor'
- Fixing typo in commit message for
'usb: hub: Fix enumration timeout'
- Removing separate definition for 'structusb_ep_desc'; thereby adding 'structusb_ss_ep_comp_descriptor' to 'structusb_interface' only. As a result modifying the patch accordingly also dropping the patch 'usb: eth: Fix for updated usb interface descriptor structure'
- Dropping the patch 'usb: hub: Increase device enumeration timeout for
broken drives' for now (will come back with a solution at later point of time).
Applied all but the last one, let's now base all subsequent patches on u-boot- usb/next please.
Best regards,
Marek Vasut
v3 cover usb: ss: Some fixes and cleanup for USB super-speed support
Hi,
Quoted 36 lines · Marek Vasut
Dear Vivek Gautam,
Quoted 17 lines
Based on 'u-boot-usb' master branch.
This patch-series includes majorly some clean-up, few fixes and then some basic super-speed usb infrastructure addition, to help put support for XHCI in near future.
Changes from v2:
- Added a patch "usb: common: Weed out USB_**_PRINTFs from usb framework" to replace USB_***_PRINTF withdebug()so as to get rid of unnecessary DEBUG macros in usb common framework and thereby rebasing other patches on top of this so that noUSB_PRINTFappears further.
- Added a patch to reset only USB 2.0 ports, since USB 3.0 protocol ports don't require a reset to happen to move to 'enabled' state.
- Added a patch to move definition of 'min3()' out of ehci-hcd and putting the same as macro definition in common header.
- Using a 'switch-case' inportspeed()incmd_usb.cQuoted 16 lines
Changes from v1:
- Fixing the issue with 'ifno' as well as added 'if_desc'.
- Instead of turning-on only powered-off hub ports, power-cycle all the hub ports (aka: turning off and then turning on again power to all the ports.
- Fixing commenting style problem in
'usb: Update device class in usb device's descriptor'
- Fixing typo in commit message for
'usb: hub: Fix enumration timeout'
- Removing separate definition for 'structusb_ep_desc'; thereby adding 'structusb_ss_ep_comp_descriptor' to 'structusb_interface' only. As a result modifying the patch accordingly also dropping the patch 'usb: eth: Fix for updated usb interface descriptor structure'
- Dropping the patch 'usb: hub: Increase device enumeration timeout for
broken drives' for now (will come back with a solution at later point of time).
Quoted 2 lines
Applied all but the last one, let's now base all subsequent patches on u-boot- usb/next please.
And i can't see these patches on u-boot-usb/next. Possibly you haven't pushed out yet ??
Signature · 2 lines
Thanks & Regards Vivek
v3 cover usb: ss: Some fixes and cleanup for USB super-speed support
Dear Vivek Gautam,
Quoted 55 lines
Hi,
Quoted 52 lines · Marek Vasut
Dear Vivek Gautam,
Quoted 50 lines
Based on 'u-boot-usb' master branch.
This patch-series includes majorly some clean-up, few fixes and then some basic super-speed usb infrastructure addition, to help put support for XHCI in near future.
Changes from v2:
- Added a patch "usb: common: Weed out USB_**_PRINTFs from usb framework"to replace USB_***_PRINTF with
debug()so as to get rid of unnecessary DEBUG macros in usb common framework and thereby rebasing other patches on top of this so that noUSB_PRINTFappears further.- Added a patch to reset only USB 2.0 ports, since USB 3.0 protocol ports
don't require a reset to happen to move to 'enabled' state.
- Added a patch to move definition of '
min3()' out of ehci-hcd and puttingthe same as macro definition in common header.
- Using a 'switch-case' in
portspeed()incmd_usb.cChanges from v1:
- Fixing the issue with 'ifno' as well as added 'if_desc'.
- Instead of turning-on only powered-off hub ports, power-cycle allthe hub ports (aka: turning off and then turning on again power to all the ports.
- Fixing commenting style problem in
'usb: Update device class in usb device's descriptor'
- Fixing typo in commit message for
'usb: hub: Fix enumration timeout'
- Removing separate definition for 'struct
usb_ep_desc'; thereby adding'struct
usb_ss_ep_comp_descriptor' to 'structusb_interface' only. As a result modifying the patch accordingly also dropping the patch 'usb: eth: Fix for updated usb interface descriptor structure'- Dropping the patch 'usb: hub: Increase device enumeration timeout for
Quoted 8 lines
Quoted 5 lines
Quoted 2 lines
broken drives' for now (will come back with a solution at later point of time).
Applied all but the last one, let's now base all subsequent patches on u-boot- usb/next please.
And i can't see these patches on u-boot-usb/next. Possibly you haven't pushed out yet ??
You're right, sorry! Fixed now
Best regards,
Marek Vasut
v3 cover usb: ss: Some fixes and cleanup for USB super-speed support
Quoted 60 lines · Marek Vasut
Dear Vivek Gautam,
Quoted 58 lines
Hi,
Quoted 55 lines · Marek Vasut
Dear Vivek Gautam,
Quoted 52 lines
Based on 'u-boot-usb' master branch.
This patch-series includes majorly some clean-up, few fixes and then some basic super-speed usb infrastructure addition, to help put support for XHCI in near future.
Changes from v2:
- Added a patch "usb: common: Weed out USB_**_PRINTFs from usb framework"to replace USB_***_PRINTF with
debug()so as to get rid of unnecessary DEBUG macros in usb common framework and thereby rebasing other patches on top of this so that noUSB_PRINTFappears further.- Added a patch to reset only USB 2.0 ports, since USB 3.0 protocol ports
don't require a reset to happen to move to 'enabled' state.
- Added a patch to move definition of '
min3()' out of ehci-hcd and puttingthe same as macro definition in common header.
- Using a 'switch-case' in
portspeed()incmd_usb.cChanges from v1:
- Fixing the issue with 'ifno' as well as added 'if_desc'.
- Instead of turning-on only powered-off hub ports, power-cycle allthe hub ports (aka: turning off and then turning on again power to all the ports.
- Fixing commenting style problem in
'usb: Update device class in usb device's descriptor'
- Fixing typo in commit message for
'usb: hub: Fix enumration timeout'
- Removing separate definition for 'struct
usb_ep_desc'; thereby adding'struct
usb_ss_ep_comp_descriptor' to 'structusb_interface' only. As a result modifying the patch accordingly also dropping the patch 'usb: eth: Fix for updated usb interface descriptor structure'- Dropping the patch 'usb: hub: Increase device enumeration timeout for
broken drives' for now (will come back with a solution at later point of time).
Quoted 7 lines
Quoted 5 lines
Quoted 2 lines
Applied all but the last one, let's now base all subsequent patches on u-boot- usb/next please.
And i can't see these patches on u-boot-usb/next. Possibly you haven't pushed out yet ??
You're right, sorry! Fixed now
Thanks you so much !!
Signature · 2 lines
Thanks & Regards Vivek
v3 cover usb: ss: Some fixes and cleanup for USB super-speed support
In reply to Vivek GautamDear Vivek Gautam,
Quoted 5 lines
Based on 'u-boot-usb' master branch.
This patch-series includes majorly some clean-up, few fixes and then some basic super-speed usb infrastructure addition, to help put support for XHCI in near future.
btw can you test your patches with MAKEALL -a arm? I get this:
--------------------- SUMMARY ----------------------------
Boards compiled: 306
Boards with errors: 65 ( qong mx35pdk gplugd at91sam9m10g45ek_nandflash pogo_e02 dns325 iconnect lschlv2 lsxhl d2net_v2 inetspace_v2 net2big_v2 netspace_lite_v2 netspace_max_v2 netspace_v2 wireless_space dreamplug guruplug mv88f6281gtw_ge openrd_base openrd_client openrd_ultimate rd6281a sheevaplug ib62x0 dockstar tk71 zmx25 mx23_olinuxino apx4devkit mx23evk m28evk mx28evk sc_sps_1 edminiv2 mx51_efikamx mx51_efikasb mx51evk mx53loco mx6qsabreauto mx6qsabrelite nitrogen6dl nitrogen6dl2g nitrogen6q nitrogen6q2g nitrogen6s nitrogen6s1g cm_t35 mt_ventoux omap3_beagle mcx twister omap4_panda snow smdk5250 harmony seaboard ventana whistler colibri_t20_iris plutux medcom-wide tec paz00 trimslice )
----------------------------------------------------------
Best regards,
Marek Vasut
v3 cover usb: ss: Some fixes and cleanup for USB super-speed support
Hi Marek,
Quoted 10 lines · Marek Vasut
Dear Vivek Gautam,
Quoted 5 lines
Based on 'u-boot-usb' master branch.
This patch-series includes majorly some clean-up, few fixes and then some basic super-speed usb infrastructure addition, to help put support for XHCI in near future.
btw can you test your patches with MAKEALL -a arm? I get this:
Quoted 12 lines
--------------------- SUMMARY ----------------------------
Boards compiled: 306
Boards with errors: 65 ( qong mx35pdk gplugdat91sam9m10g45ek_nandflashpogo_e02dns325 iconnect lschlv2 lsxhld2net_v2inetspace_v2net2big_v2netspace_lite_v2netspace_max_v2netspace_v2wireless_spacedreamplug guruplugmv88f6281gtw_geopenrd_baseopenrd_clientopenrd_ultimaterd6281a sheevaplug ib62x0 dockstar tk71 zmx25mx23_olinuxinoapx4devkit mx23evk m28evk mx28evksc_sps_1edminiv2mx51_efikamxmx51_efikasbmx51evk mx53loco mx6qsabreauto mx6qsabrelite nitrogen6dl nitrogen6dl2g nitrogen6q nitrogen6q2g nitrogen6s nitrogen6s1gcm_t35mt_ventouxomap3_beaglemcx twisteromap4_pandasnow smdk5250 harmony seaboard ventana whistlercolibri_t20_irisplutux medcom-wide tec paz00 trimslice )
----------------------------------------------------------
Tried with MAKEALL
got following result
--------------------- SUMMARY ---------------------------- Boards compiled: 306
Boards with errors: 1 ( omap3_evm )
Boards with warnings but no errors: 17 ( VCMA9 smdk2410 kzm9g balloon3 h2200 lubbock palmld palmtc polaris pxa255_idp trizepsiv vpac270_nor_128 vpac270_nor_256 vpac270_ond_256 xaeniax zipitz2 colibri_pxa270 )
----------------------------------------------------------
There's something to do with Cross Compiler version ?? btw what environment are you compiling the source with.
Signature · 2 lines
Thanks & Regards Vivek
v3 cover usb: ss: Some fixes and cleanup for USB super-speed support
Quoted 30 lines · Vivek Gautam
Hi Marek,
Quoted 10 lines · Marek Vasut
Dear Vivek Gautam,
Quoted 5 lines
Based on 'u-boot-usb' master branch.
This patch-series includes majorly some clean-up, few fixes and then some basic super-speed usb infrastructure addition, to help put support for XHCI in near future.
btw can you test your patches with MAKEALL -a arm? I get this:
Quoted 12 lines
--------------------- SUMMARY ----------------------------
Boards compiled: 306
Boards with errors: 65 ( qong mx35pdk gplugdat91sam9m10g45ek_nandflashpogo_e02dns325 iconnect lschlv2 lsxhld2net_v2inetspace_v2net2big_v2netspace_lite_v2netspace_max_v2netspace_v2wireless_spacedreamplug guruplugmv88f6281gtw_geopenrd_baseopenrd_clientopenrd_ultimaterd6281a sheevaplug ib62x0 dockstar tk71 zmx25mx23_olinuxinoapx4devkit mx23evk m28evk mx28evksc_sps_1edminiv2mx51_efikamxmx51_efikasbmx51evk mx53loco mx6qsabreauto mx6qsabrelite nitrogen6dl nitrogen6dl2g nitrogen6q nitrogen6q2g nitrogen6s nitrogen6s1gcm_t35mt_ventouxomap3_beaglemcx twisteromap4_pandasnow smdk5250 harmony seaboard ventana whistlercolibri_t20_irisplutux medcom-wide tec paz00 trimslice )
----------------------------------------------------------Tried with MAKEALL
got following result
Quoted 8 lines
--------------------- SUMMARY ---------------------------- Boards compiled: 306
Boards with errors: 1 (omap3_evm)
Boards with warnings but no errors: 17 ( VCMA9 smdk2410 kzm9g balloon3 h2200 lubbock palmld palmtc polarispxa255_idptrizepsivvpac270_nor_128vpac270_nor_256vpac270_ond_256xaeniax zipitz2colibri_pxa270)
----------------------------------------------------------
**Without my patches i get following result
--------------------- SUMMARY ---------------------------- Boards compiled: 306
Boards with warnings but no errors: 17 ( VCMA9 smdk2410 kzm9g balloon3 h2200 lubbock palmld palmtc polaris pxa255_idp trizepsiv vpac270_nor_128 vpac270_nor_256 vpac270_ond_256 xaeniax zipitz2 colibri_pxa270 )
----------------------------------------------------------
Quoted 3 lines
There's something to do with Cross Compiler version ?? btw what environment are you compiling the source with.
Signature · 2 lines
Thanks & Regards Vivek
v3 cover usb: ss: Some fixes and cleanup for USB super-speed support
HI Marek,
Quoted 32 lines · Vivek Gautam
Quoted 30 lines · Vivek Gautam
Hi Marek,
Quoted 10 lines · Marek Vasut
Dear Vivek Gautam,
Quoted 5 lines
Based on 'u-boot-usb' master branch.
This patch-series includes majorly some clean-up, few fixes and then some basic super-speed usb infrastructure addition, to help put support for XHCI in near future.
btw can you test your patches with MAKEALL -a arm? I get this:
Quoted 12 lines
--------------------- SUMMARY ----------------------------
Boards compiled: 306
Boards with errors: 65 ( qong mx35pdk gplugdat91sam9m10g45ek_nandflashpogo_e02dns325 iconnect lschlv2 lsxhld2net_v2inetspace_v2net2big_v2netspace_lite_v2netspace_max_v2netspace_v2wireless_spacedreamplug guruplugmv88f6281gtw_geopenrd_baseopenrd_clientopenrd_ultimaterd6281a sheevaplug ib62x0 dockstar tk71 zmx25mx23_olinuxinoapx4devkit mx23evk m28evk mx28evksc_sps_1edminiv2mx51_efikamxmx51_efikasbmx51evk mx53loco mx6qsabreauto mx6qsabrelite nitrogen6dl nitrogen6dl2g nitrogen6q nitrogen6q2g nitrogen6s nitrogen6s1gcm_t35mt_ventouxomap3_beaglemcx twisteromap4_pandasnow smdk5250 harmony seaboard ventana whistlercolibri_t20_irisplutux medcom-wide tec paz00 trimslice )
----------------------------------------------------------Tried with MAKEALL
got following result
Quoted 8 lines
Quoted 8 lines
--------------------- SUMMARY ---------------------------- Boards compiled: 306
Boards with errors: 1 (omap3_evm)
Boards with warnings but no errors: 17 ( VCMA9 smdk2410 kzm9g balloon3 h2200 lubbock palmld palmtc polarispxa255_idptrizepsivvpac270_nor_128vpac270_nor_256vpac270_ond_256xaeniax zipitz2colibri_pxa270)
----------------------------------------------------------
I actually checked now just for omap3_evm configuration by trying out:
make distclean
make omap3_evm_config
make
But strangely i didn't get any build errros for omap3_evm board on explicitly compiling for it.
Any clue ?
Quoted 4 lines
**Without my patches i get following result
Quoted 11 lines
--------------------- SUMMARY ---------------------------- Boards compiled: 306
Boards with warnings but no errors: 17 ( VCMA9 smdk2410 kzm9g balloon3 h2200 lubbock palmld palmtc polarispxa255_idptrizepsivvpac270_nor_128vpac270_nor_256vpac270_ond_256xaeniax zipitz2colibri_pxa270)
----------------------------------------------------------Quoted 3 lines
There's something to do with Cross Compiler version ?? btw what environment are you compiling the source with.
I am using "arm-2011.09" cross toolchain.
Signature · 2 lines
Thanks & Regards Vivek
v3 cover usb: ss: Some fixes and cleanup for USB super-speed support
Dear Vivek Gautam,
Quoted 3 lines
HI Marek,
On Thu, Apr 18, 2013 at 4:29 PM, Vivek Gautam <[email protected]>
Quoted 62 lines · wrote:
Quoted 42 lines
On Thu, Apr 18, 2013 at 11:54 AM, Vivek Gautam
Quoted 39 lines · <[email protected]> wrote:
Hi Marek,
Quoted 10 lines · Marek Vasut
Dear Vivek Gautam,
Quoted 5 lines
Based on 'u-boot-usb' master branch.
This patch-series includes majorly some clean-up, few fixes and then some basic super-speed usb infrastructure addition, to help put support for XHCI in near future.
btw can you test your patches with MAKEALL -a arm? I get this:
Quoted 14 lines
--------------------- SUMMARY ---------------------------- Boards compiled: 306
Boards with errors: 65 ( qong mx35pdk gplugdat91sam9m10g45ek_nandflashpogo_e02dns325 iconnect lschlv2 lsxhld2net_v2inetspace_v2net2big_v2netspace_lite_v2netspace_max_v2netspace_v2wireless_spacedreamplug guruplugmv88f6281gtw_geopenrd_baseopenrd_clientopenrd_ultimaterd6281a sheevaplug ib62x0 dockstar tk71 zmx25mx23_olinuxinoapx4devkit mx23evk m28evk mx28evksc_sps_1edminiv2mx51_efikamxmx51_efikasbmx51evk mx53loco mx6qsabreauto mx6qsabrelite nitrogen6dl nitrogen6dl2g nitrogen6q nitrogen6q2g nitrogen6s nitrogen6s1gcm_t35mt_ventouxomap3_beaglemcx twisteromap4_pandasnow smdk5250 harmony seaboard ventana whistlercolibri_t20_irisplutux medcom-wide tec paz00 trimslice )
----------------------------------------------------------Tried with MAKEALL
got following result--------------------- SUMMARY ---------------------------- Boards compiled: 306
Boards with errors: 1 (omap3_evm)
Boards with warnings but no errors: 17 ( VCMA9 smdk2410 kzm9g balloon3 h2200 lubbock palmld palmtc polarispxa255_idptrizepsivvpac270_nor_128vpac270_nor_256vpac270_ond_256xaeniax zipitz2colibri_pxa270)
----------------------------------------------------------I actually checked now just for
omap3_evmconfiguration by trying out:
make distclean
makeomap3_evm_config
makeBut strangely i didn't get any build errros for
omap3_evmboard on explicitly compiling for it.
Any clue ?Quoted 10 lines
**Without my patches i get following result
--------------------- SUMMARY ---------------------------- Boards compiled: 306
Boards with warnings but no errors: 17 ( VCMA9 smdk2410 kzm9g balloon3 h2200 lubbock palmld palmtc polarispxa255_idptrizepsivvpac270_nor_128vpac270_nor_256vpac270_ond_256xaeniax zipitz2colibri_pxa270)
----------------------------------------------------------
Quoted 4 lines
Quoted 2 lines
Quoted 2 lines
There's something to do with Cross Compiler version ?? btw what environment are you compiling the source with.
I am using "arm-2011.09" cross toolchain.
I use ELDK 5.3 and Debian 4.7.2-5 to do by builds. But now that I'm looking at it, it's this patch that caused it, sorry.
commit 28b31a5937b89528c40df24dd6c9122578880605
Author: Julius Werner <[email protected]> Date: Thu Feb 28 18:08:40 2013 +0000
usb: Add new command to set USB 2.0 port test modes
drivers/usb/host/ehci-hcd.c +0 -0
@@ -630,7 +630,7 @@ ehci_submit_root(struct usb_device *dev, unsigned long pipe,
void *buffer,
printf("The request port(%d) is not configured\n", port - 1);
return -1;
}
- status_reg = (uint32_t *)&hcor->or_portsc[port - 1];
+ status_reg = (uint32_t *)&ctrl->hcor->or_portsc[port - 1];
srclen = 0;
debug("req=%u (%#x), type=%u (%#x), value=%u, index=%u\n",This change fixes is, right Julius ?
Best regards,
Marek Vasut
v3 cover usb: ss: Some fixes and cleanup for USB super-speed support
Hi Marek,
I'm sorry, that must have slipped by when I ported the change from my local fork. Your patch is correct, you just need to add the "ctrl->" there.
Quoted 101 lines · Marek Vasut
Dear Vivek Gautam,
Quoted 3 lines
HI Marek,
On Thu, Apr 18, 2013 at 4:29 PM, Vivek Gautam <[email protected]>
Quoted 62 lines · wrote:
Quoted 42 lines
On Thu, Apr 18, 2013 at 11:54 AM, Vivek Gautam
Quoted 39 lines · <[email protected]> wrote:
Hi Marek,
Quoted 10 lines · Marek Vasut
Dear Vivek Gautam,
Quoted 5 lines
Based on 'u-boot-usb' master branch.
This patch-series includes majorly some clean-up, few fixes and then some basic super-speed usb infrastructure addition, to help put support for XHCI in near future.
btw can you test your patches with MAKEALL -a arm? I get this:
Quoted 14 lines
--------------------- SUMMARY ---------------------------- Boards compiled: 306
Boards with errors: 65 ( qong mx35pdk gplugdat91sam9m10g45ek_nandflashpogo_e02dns325 iconnect lschlv2 lsxhld2net_v2inetspace_v2net2big_v2netspace_lite_v2netspace_max_v2netspace_v2wireless_spacedreamplug guruplugmv88f6281gtw_geopenrd_baseopenrd_clientopenrd_ultimaterd6281a sheevaplug ib62x0 dockstar tk71 zmx25mx23_olinuxinoapx4devkit mx23evk m28evk mx28evksc_sps_1edminiv2mx51_efikamxmx51_efikasbmx51evk mx53loco mx6qsabreauto mx6qsabrelite nitrogen6dl nitrogen6dl2g nitrogen6q nitrogen6q2g nitrogen6s nitrogen6s1gcm_t35mt_ventouxomap3_beaglemcx twisteromap4_pandasnow smdk5250 harmony seaboard ventana whistlercolibri_t20_irisplutux medcom-wide tec paz00 trimslice )
----------------------------------------------------------Tried with MAKEALL
got following result--------------------- SUMMARY ---------------------------- Boards compiled: 306
Boards with errors: 1 (omap3_evm)
Boards with warnings but no errors: 17 ( VCMA9 smdk2410 kzm9g balloon3 h2200 lubbock palmld palmtc polarispxa255_idptrizepsivvpac270_nor_128vpac270_nor_256vpac270_ond_256xaeniax zipitz2colibri_pxa270)
----------------------------------------------------------I actually checked now just for
omap3_evmconfiguration by trying out:
make distclean
makeomap3_evm_config
makeBut strangely i didn't get any build errros for
omap3_evmboard on explicitly compiling for it.
Any clue ?Quoted 10 lines
**Without my patches i get following result
--------------------- SUMMARY ---------------------------- Boards compiled: 306
Boards with warnings but no errors: 17 ( VCMA9 smdk2410 kzm9g balloon3 h2200 lubbock palmld palmtc polarispxa255_idptrizepsivvpac270_nor_128vpac270_nor_256vpac270_ond_256xaeniax zipitz2colibri_pxa270)
----------------------------------------------------------Quoted 4 lines
Quoted 2 lines
Quoted 2 lines
There's something to do with Cross Compiler version ?? btw what environment are you compiling the source with.
I am using "arm-2011.09" cross toolchain.
I use ELDK 5.3 and Debian 4.7.2-5 to do by builds. But now that I'm looking at it, it's this patch that caused it, sorry.
commit
28b31a5937b89528c40df24dd6c9122578880605Author: Julius Werner <[email protected]> Date: Thu Feb 28 18:08:40 2013 +0000usb: Add new command to set USB 2.0 port test modes
drivers/usb/host/ehci-hcd.c +1
-1@@ -630,7 +630,7 @@ ehci_submit_root(struct usb_device *dev, unsigned long pipe, void *buffer, printf("The request port(%d) is not configured\n", port - 1);return -1 ; } - status_reg = (uint32_t *)&hcor->or_portsc[port -1 ]; + status_reg = (uint32_t *)&ctrl->hcor->or_portsc[port -1 ]; srclen =0 ;debug("req=%u (%#x), type=%u (%#x), value=%u, index=%u\n",This change fixes is, right Julius ?
Best regards,
Marek Vasut
v3 cover usb: ss: Some fixes and cleanup for USB super-speed support
Hi Julius,
Quoted 5 lines
Hi Marek,
I'm sorry, that must have slipped by when I ported the change from my local fork. Your patch is correct, you just need to add the "ctrl->" there.
Well, next time please make sure to compile-test your patches at least before you send them.
The stuff is now pushed, Vivek.
Best regards,
Marek Vasut
v3 cover usb: ss: Some fixes and cleanup for USB super-speed support
Quoted 4 lines · Marek Vasut
Hi Julius,
Quoted 2 lines
Hi Marek,
Quoted 8 lines
Quoted 3 lines
I'm sorry, that must have slipped by when I ported the change from my local fork. Your patch is correct, you just need to add the "ctrl->" there.
Well, next time please make sure to compile-test your patches at least before you send them.
The stuff is now pushed, Vivek.
Thanks Marek.
Signature · 2 lines
Best Regards Vivek