Skip to contentSkip to conversation

cw1200 and older kernels (newbie question)

DISCUSSION Solomon Peachy 3 participants 3 messages Jul 20, 2013 – Jul 23, 2013

lore
Share & tools
lore
#0 lore raw

As part of the cw1200 mainlining effort, I stripped two chunks out of the cw1200 driver that only had relevance relevance for older kernels:

1) Workaround for a hardware SDIO bug (needed for < 3.2)
* Formal SDIO-level quirk mainlined in 3.2-rc
* Absence results in device hang with 512-byte SDIO transfers.
2) Support for kernels < 2.6.36
* Driver utilizes the new workqueue API introduced in 2.6.36, taking advantage of new semantics/features (WQ_HIGHPRI + WQ_CPU_INTENSIVE)
* With older kernels we need a dedicated kthread

Backports has an implementation of the new workqueue API, but it is expressed in terms of the old API, which is inadequate for the cw1200's needs.

Perhaps it's a lack of sufficient sleep, but I haven't quite wrapped my head around how the bakports stuff is put together; it's radically different that compat-wireless, which had nice orderly patch directories. :)

Any advice on how I should proceed here? I want to do this the RightWay, but I don't know what that is yet -- Should I just add these two patches to the backports tree under patches/unified-drivers?

(The code chunks in question are wrapped with LINUX_VERSION_CODE tests,
so they function fine on newer kernels)

- Solomon

Signature · 3 lines
Solomon Peachy        		       pizza at shaftnet dot org	 
Delray Beach, FL                          ^^ (email/xmpp) ^^
Quidquid latine dictum sit, altum viditur.
#1 lore raw

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Quoted 25 lines · Solomon Peachy

As part of the cw1200 mainlining effort, I stripped two chunks out of the cw1200 driver that only had relevance relevance for older kernels:

1) Workaround for a hardware SDIO bug (needed for < 3.2) * Formal SDIO-level quirk mainlined in 3.2-rc * Absence results in device hang with 512-byte SDIO transfers. 2) Support for kernels < 2.6.36
* Driver utilizes the new workqueue API introduced in 2.6.36, taking advantage of new semantics/features (WQ_HIGHPRI + WQ_CPU_INTENSIVE) * With older kernels we need a dedicated kthread

Backports has an implementation of the new workqueue API, but it is
expressed in terms of the old API, which is inadequate for the
cw1200's needs.

Perhaps it's a lack of sufficient sleep, but I haven't quite wrapped my head around how the bakports stuff is put together; it's radically different that compat-wireless, which had nice orderly patch directories. :)

Any advice on how I should proceed here? I want to do this the RightWay, but I don't know what that is yet -- Should I just add these two patches to the backports tree under patches/unified-drivers?

Quoted 5 lines

(The code chunks in question are wrapped with LINUX_VERSION_CODE tests, so they function fine on newer kernels)

- Solomon

Hi Solomon,

We have a documentation for that in the wiki [0]. Just ask if you have any further questions. There are some pending changes regarding the workqueue API changes in the kernel, you should base your work on top of that [1].

Hauke

[0]:
https://backports.wiki.kernel.org/index.php/Documentation/backports/hacking
[1]: http://www.spinics.net/lists/backports/msg01505.html
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with undefined - http://www.enigmail.net/

iQIcBAEBAgAGBQJR6pt7AAoJEIZ0px9YPRMyTv4P/0cWlLLx0lT3Ocdc1GR0y5R+ CNkOQXJTQPl2ft8yCD5kpuUF2NylknD8GX+wVreDf8BscTauSBC4yw5o3OWp5QAC t3gYYxeBupJvAjbWuJ61JFe5b9wTOM+Vc2cPBFdJoMGCy/iEc7cRkAJDwQMKnMnS N6rEtIvxk+tJjGb6UgagLPDGpwVxu+XCKPIAabWB4M/8oZjz+eGQXm9JqveP6Q2Q KsWJvPeq1CwEVFHeVAiVEF6TlV5sF0h+CivMegjMfGnRWB2ojLnNDnBv9DT0pVJz dsPXgHvoKJNpU5oMguFWuHmvUaljnjBxXvXhHIzIYsmEMeRVIprPaQKVk9B9wSER W/GWkQOgKlhpNYgFe++6Ktt3dVb2omm4mLfeWgGK8RoWXpH1skp9IMyzt22vrRON n2ASW1E7gcnAYWhEl/WjGWSyMdqfV0LZF2forJhsBA9GhcdoEqdedMQWzh7lRHOb OGgj5zhihWJiw69IiBUNpmtIGTw9YYRLFbPUxNf/WBwlcSZFpguXyAfB6G7x5UFw 90PSxg0nTsoT6DWQDSzORLucAm571sHFkwpyj6gBO+qEcZnk8XKx8KjfvM9AQZIl 4QL1wDZK5JrWTkpPSUgJTINUfrOVZTDZAMKsbOdWpeFEuu/fsghA5dyA45nPZgGw ZRhflmc8ulHQ46KryJiC
=rPy0
-----END PGP SIGNATURE-----

#2 lore raw
Quoted 3 lines · Hauke Mehrtens

There are some pending changes regarding the
workqueue API changes in the kernel, you should base your work on top of that [1].

This should be merged now with two line additions I made. Further changes / enhancements to this are welcomed however.

Luis