[procps] Re: ps - wchanf option
- From: Craig Small <csmall@xxxxxxxxxx>
- To: Jan Rybar <jrybar@xxxxxxxxxx>, procps@xxxxxxxxxxxxx, Jim Warner <james.warner@xxxxxxxxxxx>
- Date: Thu, 10 Nov 2016 20:05:30 +0000
On Fri, Nov 11, 2016 at 2:28 AM Jan Rybar <jrybar@xxxxxxxxxx> wrote:
JBD2 contains jbd2_journal_get_write_access(handle_t *, struct
buffer_head *) which calls do_get_write_access(handle_t
*handle, struct journal_head *jh, int force_copy), defined in
fs/jbd2/transaction.c.
Meanwhile include/linux/fs.h defines get_write_access(struct inode
*inode), used in e.g. fs/namei.c.
Right now we are getting somewhere. The report was saying
jbd2_journal_get_write_access was being stripped. It's not but the
underlying call is and causing confusion.
If someone sees get_write_access in ps, it is not clear for the user,
whether jbd2 block or fs inode handler is meant.
This is the detail required. You'd get get_write_access and have no idea if
it is the linux/fs.h one or the stripped do_get_write_access called by the
jbd2 function.
Technically speaking, I think the jbd2 folk did the wrong thing here, but
hey that genie is long gone out of the bottle, departed and lives in a
condo by Florida by now.
So yes we definitely need to not strip "do_" and while we are at it, might
as well remove the "sys_" stripping too because I'm sure someone somewhere
has done something like jbd2 people did. I think this full name detail
should be the default.
I still think removing the _* is a good idea because it looks like noise,
as long as it is documented.
Thanks Jan for looking into it deeper and finding the real problem.
Comments?
- Craig
--
Craig Small (@smallsees)
http://dropbear.xyz/ csmall at : enc.com.au
Debian GNU/Linux
http://www.debian.org/ csmall at : debian.org
GPG fingerprint: 5D2F B320 B825 D939 04D2 0519 3938 F96B DF50 FEA5
Other related posts:
