Changeset a94cbfa in mainline for uspace/drv/bus/usb/ehci/ehci_rh.h

Timestamp:

2018-01-26T00:55:38Z (6 years ago)

Author:

Ondřej Hlavatý <aearsis@…>

Branches:

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

Children:

77733a9

Parents:

047fbc8

git-author:

Ondřej Hlavatý <aearsis@…> (2018-01-26 00:15:17)

git-committer:

Ondřej Hlavatý <aearsis@…> (2018-01-26 00:55:38)

Message:

ehci+ohci: proper handling of unfinished transfer

Because transfers can be aborted, the RH may not store a pointer to
them. Also, it's no longer true that the batch automatically locks the
endpoint, the mechanism must be used explicitly.

File:

• uspace/drv/bus/usb/ehci/ehci_rh.h (modified) (1 diff)

uspace/drv/bus/usb/ehci/ehci_rh.h

@@ -61 +61 @@
                 uint8_t rempow[STATUS_BYTES(EHCI_MAX_PORTS) * 2];
         } __attribute__((packed)) hub_descriptor;
-        /** interrupt transfer waiting for an actual interrupt to occur */
-        usb_transfer_batch_t *unfinished_interrupt_transfer;
         bool reset_flag[EHCI_MAX_PORTS];
         bool resume_flag[EHCI_MAX_PORTS];
+
+        /*
+         * This is sort of hacky, but better than duplicating functionality.
+         * We cannot simply store a pointer to a transfer in-progress, in order
+         * to allow it to be aborted. We can however store a reference to the
+         * Status Change Endpoint. Note that this is mixing two worlds together
+         * - otherwise, the RH is "a device" and have no clue about HC, apart
+         * from accessing its registers.
+         */
+        endpoint_t *status_change_endpoint;
+
 } ehci_rh_t;