2007-05-07 09:55:16

by Tilman Schmidt

[permalink] [raw]
Subject: trying to track a change to drivers/isdn/usb-gigaset.c

While looking for something else I stumbled upon an unfamiliar
change to the file drivers/isdn/usb-gigaset.c in the current
git version wrt 2.6.21. git-blame attributes it to d626f62b
which resolves to:

Author: Arnaldo Carvalho de Melo <[email protected]>
Date: Tue Mar 27 18:55:52 2007 -0300

[SK_BUFF]: Introduce skb_copy_from_linear_data{_offset}

To clearly state the intent of copying from linear sk_buffs, _offset being a
overly long variant but interesting for the sake of saving some bytes.

Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

But I cannot remember seeing that patch via the maintainer
addresses for ISDN or the Gigaset driver, and the LKML archive
doesn't seem to have it either.

Where was that patch discussed? What is its rationale?

And more generally: is there a way to get notified if someone
submits a patch that affects a file of which I am a maintainer?

Thanks,
Tilman

--
Tilman Schmidt E-Mail: [email protected]
Bonn, Germany
Diese Nachricht besteht zu 100% aus wiederverwerteten Bits.
Unge?ffnet mindestens haltbar bis: (siehe R?ckseite)


Attachments:
signature.asc (250.00 B)
OpenPGP digital signature

2007-05-07 10:41:15

by David Miller

[permalink] [raw]
Subject: Re: trying to track a change to drivers/isdn/usb-gigaset.c

From: Tilman Schmidt <[email protected]>
Date: Mon, 07 May 2007 11:55:06 +0200

[ [email protected] is the place to discuss kernel networking
issues, thanks ]

> But I cannot remember seeing that patch via the maintainer
> addresses for ISDN or the Gigaset driver, and the LKML archive
> doesn't seem to have it either.
>
> Where was that patch discussed? What is its rationale?

This was a wholesale change done across the entire tree, they were all
straightforward transformations that didn't change what the code did.

It was a change done so that we could hide the skb data buffering
details to the point where we could change the protocol header
pointers into 32-bit offsets which saves 4 bytes per such pointer per
protocol header stored in struct sk_buff.

We're not going to notify the maintainer of each of the hundreds of
source files we had to touch in order to pull this off, sorry.

These changes were posted to [email protected], they also
sat in Andrew Morton's -mm tree for quite some time.

2007-05-07 16:23:34

by Tilman Schmidt

[permalink] [raw]
Subject: Re: trying to track a change to drivers/isdn/usb-gigaset.c

David Miller schrieb:
> [ [email protected] is the place to discuss kernel networking
> issues, thanks ]

OIC. My bad. I didn't make the connection from "skb" to "networking". Sorry.

> It was a change done so that we could hide the skb data buffering
> details to the point where we could change the protocol header
> pointers into 32-bit offsets which saves 4 bytes per such pointer per
> protocol header stored in struct sk_buff.

Thanks for the explanation.

--
Tilman Schmidt E-Mail: [email protected]
Bonn, Germany
Diese Nachricht besteht zu 100% aus wiederverwerteten Bits.
Unge?ffnet mindestens haltbar bis: (siehe R?ckseite)


Attachments:
signature.asc (250.00 B)
OpenPGP digital signature