Question about "mount" command (continued)
Hello there! I moved to a new job so my old email archive is lost. I
can't quote the whole discussion, here is an original letter (found in the list archive):
--- cut ---
Also i would like to ask someone who is responsible for "mount" tool.
I'd suggest to modify it in order to support several lines in fstab with
the same device and mount points but different filesystems and options.
For example:/dev/cdrom /mnt/cdrom udf,iso9660 user,noauto,iocharset=koi8-r 0 0/dev/cdrom /mnt/cdrom hfsplus user,noauto,nls=koi8-r 0 0/dev/cdrom /mnt/cdrom hfs user,noauto,iocharset=koi8-r,codepage=10007 0 0
Currently mount will stop at the first line and produce an error if
the filesystem is not udf or iso (in the example). It will ignore the
following lines.
This would greatly improve handling of removable devices. Is there
something (standards for example) which could prevent from this
implementation?
--- cut ---
There was a suggestion to unify all the arguments in order to be able
to write something like:/dev/cdrom /mnt/cdrom hfsplus,hfs,udf,iso9660 user,noauto,iocharset=koi8-r,codepage=10007 0 0
But there is one flaw. Let's take a floppy drive as an example, where
Mac or Windows diskettes can be read. We could write:/dev/floppy /mnt/floppy hfs,vfat user,noauto,iocharset=koi8-r,codepage=10007 0 0
But there is a problem. Mac diskettes use different codepage from
Windows diskettes (they use 866). This leads to a problem. So the solution could be to write two lines instead of one:/dev/floppy /mnt/floppy vfat user,noauto,iocharset=koi8-r,codepage=866 0 0/dev/floppy /mnt/floppy hfs user,noauto,iocharset=koi8-r,codepage=10007 0 0
Also i would suggest to change argument parsing policy for filesystems
a bit. I suggest to silently ignore unknown arguments instead of producing an error. This would make autofs much more usable (i would be able to specify the whole set of needed arguments for all possible filesystems). For example something like:/dev/hdd4 /mnt/zip ext2fs,vfat user,noauto,iocharset=koi8-r,codepage=866 0 0
will not work when ext2 cartridge is inserted because ext2fs doesn't
understand "iocharset" and "codepage" and will give an error. Removing these two options produces disadvantage when reading FAT cartridge.
This would also fix problem with "user,noauto" for smbfs automatically
(i presented the patch earlier but it was rejected because "there's no need to pollute the kernel with dummy arguments, smbmount should be fixed instead").
Kind regards
Quoted 2 lines · Pavel Fedin
Hello there! I moved to a new job so my old email archive is lost. I
can't quote the whole discussion, here is an original letter (found in the list archive):
Still see no reply. Is there noone interested? Or may be i just should
throw in a patch?
I just want to save my time a bit in order not to do the work which
would be rejected (like with previous smbfs fix).
Signature · 2 lines
Kind regards, Pavel Fedin
Quoted 6 lines · Eric Van Hensbergen
You probably aren't getting much of a response as fsdevel is kernel list and mount is a user-land app. The linux version of mount is
maintained as part of the util-linux package. It's maintained on
kernel.org by Adrian Bunk <bunk__at__stusta__dot__de>. You may have better luck contacting him directly, I think there is also a mailing list just for util-linux, but I could be mistaken.
Yes, i know, i just post here because this question has close
relationship to vfs operation in general and, besides, there's a second part in the message with question about changing behavour of all filesystems so that they would ignore unknown arguments rather than raise an error and refuse to mount.
The result i would like to achieve can be obtained in several ways, so
i ask which parts should be implemented in kernel space and which ones in user space.
I ask this because two of my patches didn't follow some policies and
were rejected. I wouldn't like to produce more bad work.
Signature · 2 lines
Kind regards, Pavel Fedin