[Prism54-devel] Re: fix for islpci_dev.h

Margit Schubert-While margitsw@t-online.de
Wed, 28 Jan 2004 14:05:25 +0100


--=====================_19168212==_.ALT
Content-Type: text/plain; charset="us-ascii"; format=flowed

Hi Herbert,
At 11:06 28.01.2004 +0100, you wrote:
>On Wed, 2004-01-28 at 09:34, Margit Schubert-While wrote:
> > Luis, are we talking about the same thing ? (per subject line)
> >
> > #if LINUX_VERSION_CODE < KERNEL_VERSION(2,4,23)
> > #define free_netdev(x)  kfree(x)
> > #define pci_name(x)     x->slot_name
> > #endif
>
>well, what about adding those 4 lines in the README file, as a hint for
>people who really want to use some non-supported (i.e. they might be on
>their own if something goes wrong) kernel version, and also describe
>roughly what additional backported stuff (like firmware_class) would be
>required (which curiously happens to be part of the supported
>versions...)?

         Well, I would argue your use of "non-supported". It is the vanilla
         kernel that is "non-supported" in the sense of the word. For all
         distro's you have E-Mail/telephone support. (And, in fact, 
installing a vanilla
         kernel negates this support)
         And regarding firmware class, if you look at my post in devel
         ("help text and nitpicks"), you will see a suggested change to the 2.4
         Config.in (which is currently wrong) such that, if there is no 
firmware loader
         support in the kernel or it is not selected, then you will not get 
the choice
         for Prism.
         I really don't understand why you guys are being so pedantic about 
this.
         With the 4 line change and the Config.in changes, we are covered 
on all fronts.
         1) Kernel >= 2.4.23 OK, FW Loading if selected
         2) Kernel < 2.4.23 OK, If FW loading in kernel - OK
                                 If FW loading not in kernel or not 
selected - No Prism selection

         This is foolproof as regards the kernel patch.
         For the tarball, if one includes linux/config.h, then you can 
check that one of
         CONFIG_FW_LOADER or CONFIG_FW_LOADER_MODULE is defined.
         If not, bomb out on the compile.

         Your Linux-newbie can then hardly do anything wrong whatever 
kernel he is running.

         Cheers

         Margit  
--=====================_19168212==_.ALT
Content-Type: text/html; charset="us-ascii"

<html>
<font size=3>Hi Herbert,<br>
At 11:06 28.01.2004 +0100, you wrote:<br>
<blockquote type=cite class=cite cite>On Wed, 2004-01-28 at 09:34, Margit
Schubert-While wrote:<br>
&gt; Luis, are we talking about the same thing ? (per subject line)<br>
&gt; <br>
&gt; #if LINUX_VERSION_CODE &lt; KERNEL_VERSION(2,4,23)<br>
&gt; #define free_netdev(x)&nbsp; kfree(x)<br>
&gt; #define pci_name(x)&nbsp;&nbsp;&nbsp;&nbsp; x-&gt;slot_name<br>
&gt; #endif<br><br>
well, what about adding those 4 lines in the README file, as a hint
for<br>
people who really want to use some non-supported (i.e. they might be
on<br>
their own if something goes wrong) kernel version, and also 
describe<br>
roughly what additional backported stuff (like firmware_class) would
be<br>
required (which curiously happens to be part of the supported<br>
versions...)?</blockquote><br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>Well, I
would argue your use of &quot;non-supported&quot;. It is the 
vanilla<br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>kernel
that is &quot;non-supported&quot; in the sense of the word. For all<br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>distro's
you have E-Mail/telephone support. (And, in fact, installing a
vanilla<br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>kernel
negates this support)<br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>And
regarding firmware class, if you look at my post in devel<br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>(&quot;help
text and nitpicks&quot;), you will see a suggested change to the 
2.4<br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>Config.in
(which is currently wrong) such that, if there is no firmware 
loader<br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>support in
the kernel or it is not selected, then you will not get the choice<br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>for
Prism.<br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>I really
don't understand why you guys are being so pedantic about this.<br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>With the 4
line change and the Config.in changes, we are covered on all 
fronts.<br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>1) Kernel
&gt;= 2.4.23 OK, FW Loading if selected<br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>2) Kernel
&lt; 2.4.23 OK, If FW loading in kernel - OK<br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab><x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab><x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab><x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>If
FW loading not in kernel or not selected - No Prism selection<br><br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>This is
foolproof as regards the kernel patch.<br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>For the
tarball, if one includes linux/config.h, then you can check that one
of<br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>CONFIG_FW_LOADER
or CONFIG_FW_LOADER_MODULE is defined. <br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>If not,
bomb out on the compile.<br><br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>Your
Linux-newbie can then hardly do anything wrong whatever kernel he is
running.<br><br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>Cheers<br><br>
<x-tab>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</x-tab>Margit<x-tab>&nbsp;&nbsp;</x-tab></font></html>

--=====================_19168212==_.ALT--