Changeset 889146e in mainline for uspace/drv/bus/usb/xhci/hc.h

Timestamp:

2017-12-10T21:49:12Z (6 years ago)

Author:

Ondřej Hlavatý <aearsis@…>

Branches:

lfn, master, serial, ticket/834-toolchain-update, topic/msim-upgrade, topic/simplify-dev-export

Children:

53db806

Parents:

6ef407b

git-author:

Ondřej Hlavatý <aearsis@…> (2017-12-10 21:43:47)

git-committer:

Ondřej Hlavatý <aearsis@…> (2017-12-10 21:49:12)

Message:

xhci: commands shall not just timeout

Previous behavior was breaking semantic: if a command was successful,
but just took too long to complete, we returned an error, and the caller
had no way to know if the command's effect has taken place.

This commit implements command aborting. The wait_for_command now cannot
just timeout - instead it aborts currently running (probably blocked)
command, and then gets back to waiting. So now, if command_sync returns
an error, it means the command was really unsuccessful.

If aborting the command takes too long, we should reset the whole HC.
This is not yet implemented.

File:

• uspace/drv/bus/usb/xhci/hc.h (modified) (4 diffs)

uspace/drv/bus/usb/xhci/hc.h

@@ -45 +45 @@
 
 #include "rh.h"
+#include "commands.h"
 #include "bus.h"
+
+typedef struct xhci_command xhci_cmd_t;
 
 typedef struct xhci_hc {
@@ -61 +64 @@
 
         /* Structures in allocated memory */
-        xhci_trb_ring_t command_ring;
         xhci_event_ring_t event_ring;
         uint64_t *dcbaa;
@@ -67 +69 @@
         dma_buffer_t scratchpad_array;
         dma_buffer_t *scratchpad_buffers;
+
+        /* Command ring management */
+        xhci_cmd_ring_t cr;
 
         /* Root hub emulation */
@@ -81 +86 @@
         xhci_port_speed_t speeds [16];
         uint8_t speed_to_psiv [USB_SPEED_MAX];
-
-        /* Command list */
-        list_t commands;
-        fibril_mutex_t commands_mtx;
 
         /* TODO: Hack. Figure out a better way. */