From mboxrd@z Thu Jan 1 00:00:00 1970 From: Chris Faylor To: "'cygwin@sources.redhat.com'" , cygwin@sourceware.cygnus.com Subject: Re: Setting cygdrive prefix to '/' causes disappearing mounts Date: Tue, 19 Sep 2000 04:22:00 -0000 Message-id: <20000919072128.D27571@cygnus.com> References: <17B78BDF120BD411B70100500422FC6309E0FB@IIS000> X-SW-Source: 2000-09/msg00642.html On Tue, Sep 19, 2000 at 10:02:15AM +0200, Bernard Dautrevaux wrote: >> On Tue, Sep 19, 2000 at 03:26:27AM +0900, Kazuhiro Fujieda wrote: >>>>>>On Fri, 15 Sep 2000 13:30:27 -0400 (EDT) David Condon >>>>>>said: >>> >>>>I have been experiencing a problem in which everything in the mount >>>>table disappears. This happens when any program using the Cygwin DLL >>>>is executed. With Regedit open, I can see that the keys containing the >>>>mount points disappear. >>> >>>I've fixed this problem in the development sources. I've eliminated >>>the code deleting mount entries of which posix paths have the cygdrive >>>prefix. This fix will appear in the next release. >> >>Thanks very much for doing this. It is much appreciated. >> >>This should also be in the next snapshot. >> > >I don't really understand what's the matter exactly here; as I've >understood, the cygdrive prefix was set as "/cygdrive", so that "a:" is >available as "/cygdrive/a". Setting the cygdrive prefix as "/" should >give "a:" available as "//a", so that should *not* erase "/bin" as bin >is not a subdirectory of the virtual / subdirectory of "/" but a >subdirectory of "/" itself. No. This makes a: available as /a. >I think the matching code is wrong; what should probably be avoided is >either setting the cygdrive prefix to "", or to "/.", because then /bin >would be under cygdrive, but setting the cygdrive prefix to "/" should >*not* cause problems for "/bin"; it could however cause problems with >an UNC path to machine "X", as //X/ will try to access the local X: >drive, and that's why the convention was changed to use >(user-changeable) "/cygdrive/" prefix instead of "//". That's not how it works. That mount logic does maximum length matching, so /bin is matched first. Have you even tried this? >I'm just not sure this patch will not cause new kinds of problem; I >think fixing the matching code would be better: it should check for >mount points named /something, not "prefix>something", and avoid collapsing multiple leading "/". > >Just my 2c, Did you actually look at the patch or are you just speculating? I don't know what "matching code" you are referring to but the patch is correct. Cygwin used to umount any directory that was a subdirectory of a cygdrive path. Now it doesn't. However, if you think that something is wrong with the patch, I'm sure that both Kazuhiro and I would both love to see a correction. cgf -- Want to unsubscribe from this list? Send a message to cygwin-unsubscribe@sourceware.cygnus.com