From: Albert Lee <albertcc@tw.ibm.com>
To: Jeff Garzik <jeff@garzik.org>
Cc: Tejun Heo <htejun@gmail.com>, Alan Cox <alan@lxorguk.ukuu.org.uk>,
Linux IDE <linux-ide@vger.kernel.org>,
Doug Maxey <dwm@enoyolf.org>,
bzolnier@gmail.com, Mark Lord <liml@rtr.ca>
Subject: [PATCH 6/7] libata: push part of the irq driven pio out to workqueue
Date: Fri, 11 May 2007 15:37:57 +0800 [thread overview]
Message-ID: <46441D55.20507@tw.ibm.com> (raw)
In-Reply-To: <4644192A.8090809@tw.ibm.com>
patch 6/7:
Push part of the irq driven pio out to workqueue.
Signed-off-by: Albert Lee <albertcc@tw.ibm.com>
---
It seems this creates new race condition with ata_exec_internal_sg() and EH?
diff -Nrup 05_narrow_lock/drivers/ata/libata-core.c 06_irq_wq/drivers/ata/libata-core.c
--- 05_narrow_lock/drivers/ata/libata-core.c 2007-05-11 11:46:41.000000000 +0800
+++ 06_irq_wq/drivers/ata/libata-core.c 2007-05-11 11:54:56.000000000 +0800
@@ -4370,20 +4370,15 @@ err_out:
static inline int ata_hsm_ok_in_wq(struct ata_port *ap, struct ata_queued_cmd *qc)
{
- if (qc->tf.flags & ATA_TFLAG_POLLING)
- return 1;
-
- if (ap->hsm_task_state == HSM_ST_FIRST) {
- if (qc->tf.protocol == ATA_PROT_PIO &&
- (qc->tf.flags & ATA_TFLAG_WRITE))
- return 1;
-
- if (is_atapi_taskfile(&qc->tf) &&
- !(qc->dev->flags & ATA_DFLAG_CDB_INTR))
- return 1;
+ switch (qc->tf.protocol) {
+ case ATA_PROT_DMA:
+ return 0;
+ case ATA_PROT_ATAPI_DMA:
+ if (ap->hsm_task_state != HSM_ST_FIRST)
+ return 0;
}
- return 0;
+ return (ap->pflags & ATA_PFLAG_HSM_WQ) ? 1 : 0;
}
/**
@@ -4559,7 +4554,7 @@ fsm_start:
goto fsm_start;
}
- atapi_pio_bytes(qc, 0);
+ atapi_pio_bytes(qc, in_wq);
if (unlikely(ap->hsm_task_state == HSM_ST_ERR))
/* bad ireason reported by device */
@@ -4614,7 +4609,7 @@ fsm_start:
goto fsm_start;
}
- ata_pio_sectors(qc, 0);
+ ata_pio_sectors(qc, in_wq);
if (ap->hsm_task_state == HSM_ST_IDLE) {
/* all data read */
@@ -4713,6 +4708,31 @@ fsm_start:
goto fsm_start;
}
+static void ata_irq_task(struct work_struct *work)
+{
+ struct ata_port *ap =
+ container_of(work, struct ata_port, port_task.work);
+ struct ata_queued_cmd *qc = ap->port_task_data;
+ u8 status;
+
+ WARN_ON(ap->hsm_task_state == HSM_ST_IDLE);
+
+ /* double check the device is not busy */
+ status = ata_chk_status(ap);
+ if (status & ATA_BUSY) {
+ ata_port_printk(ap, KERN_ERR, "Unexpected device busy "
+ "(Status 0x%x)\n", status);
+ return;
+ }
+
+ /* move the HSM */
+ ata_hsm_move(ap, qc, status, 1);
+
+ /* another command or interrupt handler
+ * may be running at this point.
+ */
+}
+
/**
* ata_qc_new - Request an available ATA command, for queueing
* @ap: Port associated with device @dev
@@ -5250,11 +5270,20 @@ inline unsigned int ata_host_intr (struc
/* ack bmdma irq events */
ap->ops->irq_clear(ap);
- ata_hsm_move(ap, qc, status, 0);
+ /* invoke HSM */
+ switch (ap->hsm_task_state) {
+ case HSM_ST_FIRST:
+ case HSM_ST:
+ ap->pflags |= ATA_PFLAG_HSM_WQ;
+ ata_port_queue_task(ap, ata_irq_task, qc, 0);
+ break;
+ default:
+ ata_hsm_move(ap, qc, status, 0);
- if (unlikely(qc->err_mask) && (qc->tf.protocol == ATA_PROT_DMA ||
- qc->tf.protocol == ATA_PROT_ATAPI_DMA))
- ata_ehi_push_desc(ehi, "BMDMA stat 0x%x", host_stat);
+ if (unlikely(qc->err_mask) && (qc->tf.protocol == ATA_PROT_DMA ||
+ qc->tf.protocol == ATA_PROT_ATAPI_DMA))
+ ata_ehi_push_desc(ehi, "BMDMA stat 0x%x", host_stat);
+ }
return 1; /* irq handled */
next prev parent reply other threads:[~2007-05-11 7:38 UTC|newest]
Thread overview: 52+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-05-07 4:30 [PATCH] libata: disable_irq() during polling IDENTIFY Albert Lee
2007-05-07 7:43 ` Tejun Heo
2007-05-07 11:18 ` Alan Cox
2007-05-07 11:32 ` Tejun Heo
2007-05-08 13:36 ` Mark Lord
2007-05-07 11:19 ` Albert Lee
2007-05-07 11:29 ` Tejun Heo
2007-05-07 11:54 ` Albert Lee
2007-05-07 12:01 ` Tejun Heo
2007-05-08 11:30 ` [PATCH] libata: disable_irq() during polling IDENTIFY (take 2) Albert Lee
2007-05-08 11:41 ` Tejun Heo
2007-05-08 12:00 ` Alan Cox
2007-05-08 12:01 ` Tejun Heo
2007-05-08 12:20 ` Alan Cox
2007-05-08 12:27 ` Tejun Heo
2007-05-08 12:43 ` Alan Cox
2007-05-08 12:45 ` Tejun Heo
2007-05-08 12:45 ` Alan Cox
2007-05-08 12:57 ` Tejun Heo
2007-05-08 14:59 ` Albert Lee
2007-05-08 15:16 ` Jeff Garzik
2007-05-11 7:20 ` [PATCH/RFC 0/7] libata: push part of irq driven pio to workqueue Albert Lee
2007-05-11 7:24 ` [PATCH 1/7] libata: set the state after "PIO data-in" to HSM_ST_IDLE instead of HSM_ST_LAST Albert Lee
2007-05-11 7:28 ` [PATCH 2/7] libata: move the ata_altstatus() in ata_hsm_qc_complete() Albert Lee
2007-05-11 7:30 ` [PATCH 3/7] libata: move ata_altstatus() out to the pio data xfer functions Albert Lee
2007-05-11 7:31 ` [PATCH 4/7] libata: move polling idle irq check to ata_host_intr() Albert Lee
2007-05-11 14:27 ` Tejun Heo
2007-05-11 15:25 ` Albert Lee
2007-05-11 7:35 ` [PATCH 5/7] libata: move and reduce locking to the pio data xfer functions Albert Lee
2007-05-11 14:37 ` Tejun Heo
2007-05-11 14:55 ` Alan Cox
2007-05-11 14:57 ` Tejun Heo
2007-05-11 15:12 ` Alan Cox
2007-05-11 15:14 ` Tejun Heo
2007-05-11 15:24 ` Alan Cox
2007-05-11 15:39 ` Alan Cox
2007-05-11 16:59 ` Tejun Heo
2007-05-11 17:46 ` Alan Cox
2007-05-11 17:53 ` Tejun Heo
2007-05-11 22:00 ` Alan Cox
2007-05-14 8:24 ` Tejun Heo
2007-05-14 11:29 ` Alan Cox
2007-05-11 15:48 ` Albert Lee
2007-05-11 17:06 ` Tejun Heo
2007-05-11 17:38 ` Alan Cox
2007-05-11 17:42 ` Tejun Heo
2007-05-11 17:07 ` Tejun Heo
2007-05-11 7:37 ` Albert Lee [this message]
2007-05-11 7:41 ` [PATCH 7/7] libata: ack unexpected INTRQ when polling Albert Lee
2007-05-07 14:28 ` [PATCH] libata: disable_irq() during polling IDENTIFY Alan Cox
2007-05-08 13:42 ` Mark Lord
2007-05-08 13:57 ` Alan Cox
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=46441D55.20507@tw.ibm.com \
--to=albertcc@tw.ibm.com \
--cc=alan@lxorguk.ukuu.org.uk \
--cc=albertl@mail.com \
--cc=bzolnier@gmail.com \
--cc=dwm@enoyolf.org \
--cc=htejun@gmail.com \
--cc=jeff@garzik.org \
--cc=liml@rtr.ca \
--cc=linux-ide@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).