> From: Alan Stern [mailto:stern@xxxxxxxxxxxxxxxxxxx] > Sent: Friday, September 30, 2011 1:02 PM <snip> > Hmmm, I don't remember the full story. Probably it could be tracked > down in the email archives. > > For the record, I have a local patch in my own tree (never submitted to > Greg) that includes what you have written plus a little more, although > I didn't bother to change f_mass_storage.c along with file_storage.c. > It is given below. Differences from the current state of the source > file are at least partly explained by the fact that I haven't updated > my repository since kernel.org went down. > > Alan Stern > > > > > Index: usb-3.1/drivers/usb/gadget/file_storage.c > =================================================================== > --- usb-3.1.orig/drivers/usb/gadget/file_storage.c > +++ usb-3.1/drivers/usb/gadget/file_storage.c > @@ -1334,9 +1334,7 @@ static int do_write(struct fsg_dev *fsg) > * equal to the buffer size, which is divisible by > * the bulk-out maxpacket size. > */ > - bh->outreq->length = bh->bulk_out_intended_length = > - amount; > - bh->outreq->short_not_ok = 1; > + set_bulk_out_req_length(fsg, bh, amount); > start_transfer(fsg, fsg->bulk_out, bh->outreq, > &bh->outreq_busy, &bh->state); > fsg->next_buffhd_to_fill = bh->next; > @@ -1369,6 +1367,11 @@ static int do_write(struct fsg_dev *fsg) > amount = curlun->file_length - file_offset; > } > > + /* Don't accept excess data. The spec doesn't say > + * what to do in this case. We'll ignore the error. > + */ > + amount = min(amount, bh->bulk_out_intended_length); > + > /* Don't write a partial block */ > amount = round_down(amount, curlun->blksize); > if (amount == 0) > @@ -1408,7 +1411,7 @@ static int do_write(struct fsg_dev *fsg) > > empty_write: > /* Did the host decide to stop early? */ > - if (bh->outreq->actual != bh->outreq->length) { > + if (bh->outreq->actual < bh->bulk_out_intended_length) { > fsg->short_packet_received = 1; > break; > } > @@ -1952,7 +1955,7 @@ static int throw_away_data(struct fsg_de > fsg->next_buffhd_to_drain = bh->next; > > /* A short packet or an error ends everything */ > - if (bh->outreq->actual != bh->outreq->length || > + if (bh->outreq->actual < bh->bulk_out_intended_length || > bh->outreq->status != 0) { > raise_exception(fsg, FSG_STATE_ABORT_BULK_OUT); > return -EINTR; > @@ -1970,9 +1973,7 @@ static int throw_away_data(struct fsg_de > * equal to the buffer size, which is divisible by > * the bulk-out maxpacket size. > */ > - bh->outreq->length = bh->bulk_out_intended_length = > - amount; > - bh->outreq->short_not_ok = 1; > + set_bulk_out_req_length(fsg, bh, amount); > start_transfer(fsg, fsg->bulk_out, bh->outreq, > &bh->outreq_busy, &bh->state); > fsg->next_buffhd_to_fill = bh->next; > @@ -2651,7 +2652,6 @@ static int get_next_command(struct fsg_d > > /* Queue a request to read a Bulk-only CBW */ > set_bulk_out_req_length(fsg, bh, USB_BULK_CB_WRAP_LEN); > - bh->outreq->short_not_ok = 1; > start_transfer(fsg, fsg->bulk_out, bh->outreq, > &bh->outreq_busy, &bh->state); > Hi Alan, Yes, your patch is more correct I think, mine doesn't take care of writing the correct amount of data to the backing file. But your patch misses the code in throw_away_data() that does the same thing. Do you mind if I fix that up, make a similar patch for f_mass_storage.c, and submit it? -- Paul ��.n��������+%������w��{.n�����{���)�� jg��� �����ݢj����G���� ���j:+v���w�m������w�������h�����٥
