Hi Ray, Thank you for information. Your analysis is right. I also noticed same problem. If you'd like to submit patch, not to report, please send it with regular form. Or I could send similar patch I have.
Thanks, Seungwon Jeon On Wed, Oct 23, 2013, Ray Jui wrote: Hi Seungwon/Chris, We recently came across an eMMC issue that causes mmc_suspend to stuck waiting to claim the host. This issue only happens when one of the MMC partitions is mounted with the discard option. By tracing the issue down in the kernel MMC stack, in the function mmc_blk_issue_rq under drivers/mmc/card/block.c, we noticed the host will be released when req is valid and when req->cmd_flags is set to one of the masks: MMC_REQ_SPECIAL_MASK. Based on the code comment, I believe the intention is to always release the host in the case of special request (which include both REQ_DISCARD and REQ_FLUSH). Note in function call mmc_blk_issue_discard_rq, the memory where req points to can be freed. This means at the bottom of the mmc_blk_issue_rq, req->cmd_flags may contain garbage value and I believe this is what’s causing the issue that we are seeing in mmc_suspend after the discard operation: In mmc_suspend, it blocks waiting to claim the host but the host was never released after the discard operation. Please help to review the attached patch that fixes the issue. Thanks, Ray Jui -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
