Hi Bin,
On Wed, Sep 24, 2014 at 01:55:14PM -0500, Bin Liu wrote:
> Hi Felipe and all,
>
> The musb driver musb_host_tx() has the following:
>
> 1244 /* with CPPI, DMA sometimes triggers "extra" irqs */
> 1245 if (!urb) {
> 1246 dev_dbg(musb->controller, "extra TX%d ready, csr
> %04x\n", epnum, tx_csr);
> 1247 return;
> 1248 }
>
> and
>
> 1321 /* second cppi case */
> 1322 if (dma_channel_status(dma) == MUSB_DMA_STATUS_BUSY) {
> 1323 dev_dbg(musb->controller, "extra TX%d ready, csr
> %04x\n", epnum, tx_csr);
> 1324 return;
> 1325 }
>
> which come with the very first commit of musb driver and never got
> changed. Is there any more information about the 'extra' irqs? and
> what is the 'second cppi case'?I'm afraid the details of that were long after Dave B left us, unfortunately. He was the only one dealing with MUSB on those early phases. > I ran into this problem and musb stops working after hits here. > > [13542.933563] musb-hdrc musb-hdrc.1: qh c261ad80 urb eec92bc0 dev4 > ep7in-bulk, hw_ep 2, c2599000/4096 > [13542.953552] musb-hdrc musb-hdrc.1: usbintr (0) epintr(4) > [13542.953582] musb-hdrc musb-hdrc.1: extra TX2 ready, csr 0004 > > I bet this 'extra TX2 ready' log is printed by the first case, because > only RX2 is used in this test, and TX2 is never got used, so there > should not be any urb in TX2 queue. I am still waiting for another > test failure to confirm this, but I am trying to understand why TX2 > interrupt happened...which seems to be related RX2 request. > > Thanks for any help. unfortunately no clue on what's going on :-( -- balbi
signature.asc
Description: Digital signature
