On 25 September 2013 04:31, Haijun Zhang <Haijun.Zhang@xxxxxxxxxxxxx> wrote:
> When card is in cpu polling mode to detect card present. Card detecting
> task will be scheduled about once every second. When card is busy in large
> file transfer, detecting task will be hang and call trace will be prompt.
> When handling the request, CMD13 is always followed by every command when
> it was complete. So assume that card is present to avoid this duplicate
> detecting. Only polling card when card is free to reduce conflict with
> data transfer.
>
> <7>mmc0: req done (CMD13): 0: 00000e00 00000000 00000000 00000000
> INFO: task kworker/u:1:12 blocked for more than 120 seconds.
> "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> kworker/u:1 D 00000000 0 12 2 0x00000000
> Call Trace:
> [ee06dd50] [44042028] 0x44042028
> (unreliable)
> [ee06de10] [c0007a0c] __switch_to+0xa0/0xf0
> [ee06de30] [c04dd50c] __schedule+0x1f8/0x4a4
>
> [ee06dea0] [c04dd898] schedule+0x30/0xbc
>
> [ee06deb0] [c03816a4] __mmc_claim_host+0x98/0x19c
>
> [ee06df00] [c0385f88] mmc_sd_detect+0x38/0xc0
>
> [ee06df10] [c0382b0c] mmc_rescan+0x294/0x2e0
> [ee06df40] [c00661cc] process_one_work+0x140/0x3e0
>
> [ee06df70] [c0066bf8] worker_thread+0x18c/0x36c
> [ee06dfb0] [c006bf10] kthread+0x7c/0x80
>
> [ee06dff0] [c000de58] kernel_thread+0x4c/0x68
> <7>sdhci [sdhci_irq()]: *** mmc0 got interrupt: 0x00000001
>
> Signed-off-by: Haijun Zhang <haijun.zhang@xxxxxxxxxxxxx>
> ---
> changes for V4:
> - Enable detect change during req error recovery
>
> drivers/mmc/card/block.c | 29 +++++++++++++++++++++++++++--
> drivers/mmc/core/core.c | 5 +++++
> 2 files changed, 32 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/mmc/card/block.c b/drivers/mmc/card/block.c
> index 1a3163f..d1f1730 100644
> --- a/drivers/mmc/card/block.c
> +++ b/drivers/mmc/card/block.c
> @@ -1960,9 +1960,21 @@ static int mmc_blk_issue_rq(struct mmc_queue *mq, struct request *req)
> struct mmc_host *host = card->host;
> unsigned long flags;
>
> - if (req && !mq->mqrq_prev->req)
> + if (req && !mq->mqrq_prev->req) {
> + /*
> + * When we are here, card polling task will be blocked.
> + * So disable it to avoid this useless schedule.
> + */
> + if (host->caps & MMC_CAP_NEEDS_POLL) {
> + spin_lock_irqsave(&host->lock, flags);
> + host->rescan_disable = 1;
> + spin_unlock_irqrestore(&host->lock, flags);
> + cancel_delayed_work(&host->detect);
Sorry, I missed that you added the "cancel" operation in v3. This is
not needed, please remove.
> + }
> +
> /* claim host only for the first request */
> mmc_get_card(card);
> + }
>
> ret = mmc_blk_part_switch(card, md);
> if (ret) {
> @@ -1999,7 +2011,7 @@ static int mmc_blk_issue_rq(struct mmc_queue *mq, struct request *req)
>
> out:
> if ((!req && !(mq->flags & MMC_QUEUE_NEW_REQUEST)) ||
> - (req && (req->cmd_flags & MMC_REQ_SPECIAL_MASK)))
> + (req && (req->cmd_flags & MMC_REQ_SPECIAL_MASK))) {
> /*
> * Release host when there are no more requests
> * and after special request(discard, flush) is done.
> @@ -2007,6 +2019,19 @@ out:
> * the 'mmc_blk_issue_rq' with 'mqrq_prev->req'.
> */
> mmc_put_card(card);
> +
> + /*
> + * Detecting card status immediately in case card being
> + * removed just after the request is complete.
> + */
> + if (host->caps & MMC_CAP_NEEDS_POLL) {
> + spin_lock_irqsave(&host->lock, flags);
> + host->rescan_disable = 0;
> + spin_unlock_irqrestore(&host->lock, flags);
> + mmc_detect_change(host, 0);
> + }
> + }
> +
> return ret;
> }
>
> diff --git a/drivers/mmc/core/core.c b/drivers/mmc/core/core.c
> index bf18b6b..fea6caf 100644
> --- a/drivers/mmc/core/core.c
> +++ b/drivers/mmc/core/core.c
> @@ -2400,6 +2400,7 @@ int mmc_detect_card_removed(struct mmc_host *host)
> {
> struct mmc_card *card = host->card;
> int ret;
> + unsigned long flags;
>
> WARN_ON(!host->claimed);
>
> @@ -2414,6 +2415,10 @@ int mmc_detect_card_removed(struct mmc_host *host)
> if (!host->detect_change && !(host->caps & MMC_CAP_NEEDS_POLL))
> return ret;
>
> + spin_lock_irqsave(&host->lock, flags);
> + host->rescan_disable = 0;
> + spin_unlock_irqrestore(&host->lock, flags);
> +
This code shall remain further down, specific for polling mode when
card is detected to be removed.
So put it before "cancel_delayed_work".
> host->detect_change = 0;
> if (!ret) {
> ret = _mmc_detect_card_removed(host);
> --
> 1.8.0
>
>
Kind regards
Ulf Hansson
--
To unsubscribe from this list: send the line "unsubscribe linux-mmc" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
