James (or anyone else who can answer): The retry counters in sd.c seem to be out of whack. For example, consider sd_read_capacity(). There's an explicit retry loop that runs 3 times, and inside it is a call to scsi_execute_req() with SD_MAX_RETRIES (= 5) as an argument. If for any reason the device is not able to determine the media capacity -- other than medium not present -- the code ends up sending the READ CAPACITY command a total of 3*(5+1) = 18 times! Isn't that a little excessive? After all, if the capacity couldn't be determined the first time, how likely is the device to get it right on the eighteenth? And that's not even taking into account the fact that sd_read_capcity(), together with everything else under sd_revalidate_disk(), gets called twice in a row during device scanning. (What's the reason for that, anyway?) This makes for a total of 36 failed commands. There are other examples of explicit outer loops combined with large retry arguments to scsi_execute_req(), but this one caught my eye because of the logs attached to Bugzilla #8885. See especially the log attached to comment #22. It seems to me that we should be able to get away with using 1 instead of SD_MAX_RETRIES here, and probably in a few other places as well. There's not much point in having an inner retry loop if there's already an outer loop. Do you agree? Alan Stern - 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
