[prev in list] [next in list] [prev in thread] [next in thread] 

List:       netbsd-tech-userlevel
Subject:    Re: wedge device to name
From:       Kamil Rytarowski <n54 () gmx ! com>
Date:       2019-09-22 19:33:27
Message-ID: d02820fc-a804-9e16-c86c-c8ec31d54439 () gmx ! com
[Download RAW message or body]

[Attachment #2 (multipart/mixed)]


On 22.09.2019 19:09, Christos Zoulas wrote:
> On Sep 22,  6:59pm, n54@gmx.com (Kamil Rytarowski) wrote:
> -- Subject: Re: wedge device to name
> 
> | Looking at the diff here, just adding f_mntfromlabel[] is a better
> | choice and it does not make too much dead code in old syscall version.
> | 
> | I propose to go for versioning here.
> | 
> 
> Unfortunately the diff is deceiving. I just posted the diff that
> adds the new functionality to the code, but not the part to provide
> backwards compatibility. There 100's of lines of code in the compat
> glue just to handle the old syscalls both in libc and the kernel.
> The real issue though is what happens if you don't put the
> functionality where it belongs; look what has happened to the linux
> statvfs syscall family for example and the answers to the question
> how do I get the mountpoint of a particular file, or how do I get
> the list of mountpoints etc.
> 
> I still think that it is better to design good software interfaces
> as opposed to resorting to procfs/sysctl/ioctl or whatever glue to
> gather information even when the compat cost is high.
> 

My initial concern was caused by a misleading XML mention and threat
that this syscall will generate now XML.

The new syscall change proposal looks good and right thing to do. I have
no objections.

> christos
> 



["signature.asc" (application/pgp-signature)]

[prev in list] [next in list] [prev in thread] [next in thread]