Dan Williams <dan.j.williams@xxxxxxxxx> writes: > On Fri, Feb 3, 2012 at 8:02 PM, Eric W. Biederman <ebiederm@xxxxxxxxxxxx> wrote: >> Bjorn Helgaas <bhelgaas@xxxxxxxxxx> writes: >> >>> Eric recently added ce59791936, which is a check that complains if we >>> try to remove files from non-existent directories. >> >> Yes. It has always been a bug but since no one checked their error >> codes it has been going unnoticed. >> >>> When I remove a device (LSI SAS2008 in this case, but I don't know if >>> it's specific to that), I hit that warning three times. Let me know >>> if I can collect any more data. >> >> I tried to read that code and understand what is going on but I don't >> know enough about the block and scsi layers to have a clue where to >> start. >> >> bsg_unregister_queue seems to be a generic piece of infrastructure. >> >> q->kobj seems to be managed separately from q->bsg_dev->class_dev. >> >> If I read q->kobj correctly that is managed with blk_put_queue. >> >> So it seems like the final blk_put_queue or blk_unregister_queue is >> being called before bsg_unregister_queue and causing this problem. >> >> What the solution might be I can't believe to figure out. > > This fixed it for me: > > https://lkml.org/lkml/2012/2/8/246 It is good to see that there is a fix. But using an internal sysfs field like that is a bit scary. Eric -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
