Commit 901e3f49 authored by Julian Wiedmann's avatar Julian Wiedmann Committed by David S. Miller

s390/qeth: avoid control IO completion stalls

For control IO, qeth currently tracks the index of the buffer that it
expects to complete the next IO on each qeth_channel. If the channel
presents an IRQ while this buffer has not yet completed, no completion
processing for _any_ completed buffer takes place.
So if the 'next buffer' is skipped for any sort of reason* (eg. when it
is released due to error conditions, before the IO is started), the
buffer obviously won't switch to PROCESSED until it is eventually
allocated for a _different_ IO and completes.
Until this happens, all completion processing on that channel stalls
and pending requests possibly time out.

As a fix, remove the whole 'next buffer' logic and simply process any
IO buffer right when it completes. A channel will never have more than
one IO pending, so there's no risk of processing out-of-sequence.

*Note: currently just one location in the code really handles this problem,
       by advancing the 'next' index manually.
Signed-off-by: default avatarJulian Wiedmann <jwi@linux.vnet.ibm.com>
Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
parent 686c97ee
......@@ -557,7 +557,6 @@ enum qeth_prot_versions {
enum qeth_cmd_buffer_state {
BUF_STATE_FREE,
BUF_STATE_LOCKED,
BUF_STATE_PROCESSED,
};
enum qeth_cq {
......@@ -601,7 +600,6 @@ struct qeth_channel {
struct qeth_cmd_buffer iob[QETH_CMD_BUFFER_NO];
atomic_t irq_pending;
int io_buf_no;
int buf_no;
};
/**
......
......@@ -818,7 +818,6 @@ void qeth_clear_cmd_buffers(struct qeth_channel *channel)
for (cnt = 0; cnt < QETH_CMD_BUFFER_NO; cnt++)
qeth_release_buffer(channel, &channel->iob[cnt]);
channel->buf_no = 0;
channel->io_buf_no = 0;
}
EXPORT_SYMBOL_GPL(qeth_clear_cmd_buffers);
......@@ -924,7 +923,6 @@ static int qeth_setup_channel(struct qeth_channel *channel)
kfree(channel->iob[cnt].data);
return -ENOMEM;
}
channel->buf_no = 0;
channel->io_buf_no = 0;
atomic_set(&channel->irq_pending, 0);
spin_lock_init(&channel->iob_lock);
......@@ -1100,11 +1098,9 @@ static void qeth_irq(struct ccw_device *cdev, unsigned long intparm,
{
int rc;
int cstat, dstat;
struct qeth_cmd_buffer *buffer;
struct qeth_channel *channel;
struct qeth_card *card;
struct qeth_cmd_buffer *iob;
__u8 index;
if (__qeth_check_irb_error(cdev, intparm, irb))
return;
......@@ -1182,25 +1178,18 @@ static void qeth_irq(struct ccw_device *cdev, unsigned long intparm,
channel->state = CH_STATE_RCD_DONE;
goto out;
}
if (intparm) {
buffer = (struct qeth_cmd_buffer *) __va((addr_t)intparm);
buffer->state = BUF_STATE_PROCESSED;
}
if (channel == &card->data)
return;
if (channel == &card->read &&
channel->state == CH_STATE_UP)
__qeth_issue_next_read(card);
iob = channel->iob;
index = channel->buf_no;
while (iob[index].state == BUF_STATE_PROCESSED) {
if (iob[index].callback != NULL)
iob[index].callback(channel, iob + index);
index = (index + 1) % QETH_CMD_BUFFER_NO;
if (intparm) {
iob = (struct qeth_cmd_buffer *) __va((addr_t)intparm);
if (iob->callback)
iob->callback(iob->channel, iob);
}
channel->buf_no = index;
out:
wake_up(&card->wait_q);
return;
......@@ -2214,7 +2203,6 @@ int qeth_send_control_data(struct qeth_card *card, int len,
error:
atomic_set(&card->write.irq_pending, 0);
qeth_release_buffer(iob->channel, iob);
card->write.buf_no = (card->write.buf_no + 1) % QETH_CMD_BUFFER_NO;
rc = reply->rc;
qeth_put_reply(reply);
return rc;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment