Hi, On 04/30/2010 12:57 AM, Chris Lumens wrote:
diff --git a/storage/udev.py b/storage/udev.py index 483cd69..e3a270a 100644 --- a/storage/udev.py +++ b/storage/udev.py @@ -77,6 +77,11 @@ def udev_resolve_glob(glob): return ret def udev_get_block_devices(): + # Wait for scsi adapters to be done with scanning their busses (#583143) + iutil.execWithRedirect("modprobe", [ "scsi_wait_scan" ], + stdout = "/dev/tty5", stderr="/dev/tty5") + iutil.execWithRedirect("rmmod", [ "scsi_wait_scan" ], + stdout = "/dev/tty5", stderr="/dev/tty5") udev_settle() entries = [] for path in udev_enumerate_block_devices():So we're going to do a modprobe and an rmmod on every single call to udev_get_block_devices?
Yes, and in all those places it is appropriate to wait for any scsi adapters to be done scanning (as the existing udev_settle() call their shows).
Have you seen how many times we do that?
We do that once in the filter UI, unless an advanced target gets added, and when an advanced target gets added its appropriate to wait for adapters to be done scanning again. Then we do it a number of times in devicetree populate, where it also is appropriate in case we are doing a kickstart install, and thus have not done so in the filter ui, Last we do it in resolve_devspec, where it also is appropriate as that gets called to determine protected devices before storage gets initialized, so before other code has had a chance to do the waiting for scsi bus scanning. Regards, Hans _______________________________________________ Anaconda-devel-list mailing list Anaconda-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/anaconda-devel-list
