[BUG] 2.4.21-pre3 hdparm -X violates IDE locking

From: Ross Biro (rossb@google.com)
Date: Wed Jan 15 2003 - 13:49:38 EST


This bug impacts all versions of the kernel that I have looked at
including 2.4.18 and above. It probably impacts many other versions as
well.

hdparm -X issues a drive command ioctl which eventually issue a command
to the drive bypassing all of the normal locking. The net effect is
that a command can be set to the drive when the drive is not ready.
 Some drives lock up when this happens.

The call chain is

ide_ioctl calls
ide_cmd_ioctl

If the command is a set features to change the communications speed,
after the command is complete, ide_cmd_ioctl calls

ide_set_xfer_rate

ide_set_xfer_rate then calls drive->speedproc.

Most speedprocs then call ide_config_drive_speed
ide_config_drive_speed issue a set features command. and busy waits on it.

The easiest fix is to probably modify execute_drive_command to update
drive->current_speed when it issues the appropriate SET FEATURES
command, and then have ide_config_drive_speed do nothing if speed ==
current_speed.

I did a much bigger modification that handles changing drive speed
specially and immediately. I'll make patches for either one, but I
wanted to get some opinions before I did so.

    Ross

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Wed Jan 15 2003 - 22:00:55 EST