This is the mail archive of the
mailing list for the Cygwin project.
Re: 1.7.8-1 ls -l /proc/sys/Device causes system reset
On Wed, Mar 2, 2011 at 5:11 AM, Corinna Vinschen <address removed> wrote:
> On Mar ?2 04:23, Timothy Coalson wrote:
>> On Wed, Mar 2, 2011 at 2:40 AM, Corinna Vinschen
>> <address removed> wrote:
>> > On Mar ?1 16:41, Tim Coalson wrote:
>> >> I have found two culprits so far, SSHDRV65 and SSHDRV79. ?They seem to be
>> >> drivers related to CD copy protection software in windows.
>> > I *really* start to hate Windows XP.
>> > But I can't reproduce this, neither on XP nor on W7. ?Maybe because
>> > I don't have these SSHDRVxx devices.
>> > What puzzles me a bit is this: ?The core of the functionality is in
>> > a method called fhandler_procsys::exists(). ?This method is called
>> > even for a simple ls which does not stat. ?The underlying stat method
>> > fhandler_procsys::fstat() also just calls the exists() method and then
>> > fills the stat buffer.
>> > Hmm...
>> > There is one difference. ?If exists() is called from fstat() then
>> > there's an additional call to fetch the ACL from the device entry.
>> > That's the only additional system call in this scenario.
>> > Can you please test something for me?
>> > Fetch the WinObj tool from sysinternals:
>> > http://technet.microsoft.com/en-us/sysinternals/bb896657
>> > Install it and start it. ?In WinObj, open the Device directory and
>> > search for the SSHDRV65 and SSHDRV79 entries. ?Right-click on them and
>> > open the Properties dialog. ?Then switch to the "Security" tab.
>> > Does it work? ?Does it show something useful or a kind of error message?
>> > Corinna
>> I have switched to a different email so that I can reply properly.
>> Using the tool mentioned, both of the problem devices' properties
>> dialogs appear to be functional. ?[...]
> If I send you a link to a test Cygwin DLL via private email, would
> you mind to test it?
Sure, either email should work.
Problem reports: http://cygwin.com/problems.html
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple