PATCH: gcc3.0.2 workaround for 8139too

PATCH: gcc3.0.2 workaround for 8139too

[Ali Akcaagac]

[-- Attachment #1: Type: text/plain, Size: 1020 bytes --]

hello,

after 2.4.10 i've detected that using gcc 3.0.2 to compile the kernel
always fails with the 8139too realtek networkcard driver. errormessage
reports a bug in the compiler itself. 2.4.11-2.4.14 are affected by it
too. that's also the only file affected by this. i have investigated
into stuff and realized that there seems to be a little borkage with
understanding of strcpy. i can't tell how, why and what. i went to
irc.opensource.org and talked there with some people a while back, also
offered a 'temporarely solution' to get things compiled and operate
again (worksforme TM) and saw that a bunch of other people wanted to get
it working again in any circumstances.

so please allow me to post this 'temporarely workaround' here so other
people can profit of it until things with the gcc3.0.2 compiler are
solved.

thank you.

-- 
Name....: Ali Akcaagac
Status..: Student Of Computer & Economic Science
E-Mail..: mailto:ali.akcaagac@stud.fh-wilhelmshaven.de
WWW.....: http://www.fh-wilhelmshaven.de/~akcaagaa

[-- Attachment #2: 8139too.patch --]
[-- Type: text/x-patch, Size: 590 bytes --]

--- linux/drivers/net/8139too.c.bak	Tue Nov 20 22:39:30 2001
+++ linux/drivers/net/8139too.c	Tue Nov 20 22:39:29 2001
@@ -2393,9 +2393,9 @@
 	case ETHTOOL_GDRVINFO:
 		{
 			struct ethtool_drvinfo info = { ETHTOOL_GDRVINFO };
-			strcpy (info.driver, DRV_NAME);
-			strcpy (info.version, DRV_VERSION);
-			strcpy (info.bus_info, np->pci_dev->slot_name);
+			sprintf (info.driver, DRV_NAME);
+			sprintf (info.version, DRV_VERSION);
+			sprintf (info.bus_info, np->pci_dev->slot_name);
 			if (copy_to_user (useraddr, &info, sizeof (info)))
 				return -EFAULT;
 			return 0;

Re: PATCH: gcc3.0.2 workaround for 8139too

[Ali Akcaagac]

[-- Attachment #1: Type: text/plain, Size: 814 bytes --]

On Thu, 2001-11-22 at 03:01, Ali Akcaagac wrote:
> so please allow me to post this 'temporarely workaround' here so other
> people can profit of it until things with the gcc3.0.2 compiler are
> solved.

hello,

i've got quite a lot of response for this. oki, the problem is still
there and still no solution. somehow i was forced to do minor changes to
the previous *workaround*. at least to save myself from beeing bombed
again :) and save whoever uses this stuff, from getting complications,
please allow me to post this one.

reason: probably gcc3.0.2 bug
kernel: 2.4.10 -> X
result: solves some strange compile failure issues.

-- 
Name....: Ali Akcaagac
Status..: Student Of Computer & Economic Science
E-Mail..: mailto:ali.akcaagac@stud.fh-wilhelmshaven.de
WWW.....: http://www.fh-wilhelmshaven.de/~akcaagaa

[-- Attachment #2: 8139too.patch --]
[-- Type: text/x-patch, Size: 608 bytes --]

--- linux/drivers/net/8139too.c.bak	Tue Nov 20 22:39:30 2001
+++ linux/drivers/net/8139too.c	Sun Nov 25 11:56:49 2001
@@ -2393,9 +2393,9 @@
 	case ETHTOOL_GDRVINFO:
 		{
 			struct ethtool_drvinfo info = { ETHTOOL_GDRVINFO };
-			strcpy (info.driver, DRV_NAME);
-			strcpy (info.version, DRV_VERSION);
-			strcpy (info.bus_info, np->pci_dev->slot_name);
+			sprintf (info.driver, "%s", DRV_NAME);
+			sprintf (info.version, "%s", DRV_VERSION);
+			sprintf (info.bus_info, "%s", np->pci_dev->slot_name);
 			if (copy_to_user (useraddr, &info, sizeof (info)))
 				return -EFAULT;
 			return 0;

Re: PATCH: gcc3.0.2 workaround for 8139too

[Rene Blokland]

In article <1006691124.320.0.camel@ulixys>, Ali Akcaagac wrote:
>> so please allow me to post this 'temporarely workaround' here so other
>> people can profit of it until things with the gcc3.0.2 compiler are
>> solved.
Type make &> foo
then the internel error comes
edit foo and remove all the lines until 8139too starts.
remove the rest of the lines and the -O2 in the 8139too line.
cd d*/n*
sh ../../foo
cd /linux
make
There you go!

--
Groeten / Regards, Rene J. Blokland                                 -o)
1073KA 13 2e, the Netherlands                                       /\\
 Paai:     "Windows is als pianospelen met bokshandschoenen aan"   _\_v

Re: PATCH: gcc3.0.2 workaround for 8139too

[Ali Akcaagac]

On Sun, 2001-11-25 at 16:37, Rene Blokland wrote:
> cd d*/n*
> sh ../../foo
> cd /linux
> make
> There you go!

ok, i haven't tested this and don't even know if this works here. i for
now belive that it will work but that error has a _reason_ to happen for
some unknown issues. i'd better go for a cleaner solution right now then
forcing some crashes that _may_ happen because of borked _assembly_ code
that may be generated..

what i am thinking now (and i may be wrong). let's say i compile it your
way everything will be fine no visible errors and so on. now i use that
driver. everything works but under certain circumstances it may jump
into that CASE and crash because the compilers wasn't able to create a
clean code for it.

at least it was a quick solution to solve my personal needs that i
wanted to share with others maybe people not so involved into fixing
stuff.

but thank you. i will try this one out as soon as i update my kernel :)

-- 
Name....: Ali Akcaagac
Status..: Student Of Computer & Economic Science
E-Mail..: mailto:ali.akcaagac@stud.fh-wilhelmshaven.de
WWW.....: http://www.fh-wilhelmshaven.de/~akcaagaa

Re: PATCH: gcc3.0.2 workaround for 8139too

[Andris Pavenis]

Problem with -fomit-frame-pointer which caused gcc-3.0.X generate ICE on 
8139too.c is now fixed in gcc-3_0-branch. For example kernel 2.4.16-pre1 
builds Ok with gcc-3.0.3 20011123 (prerelease). Also 8139too.c compiles Ok 
and kernel seems to work Ok.

I have also earlier met cases when gcc-3.0.X miscompiles source (with option 
-fomit-frame-pointer only) due to same bug however it always caused crashes
of compiled source for me, not a silent data corruption.

Andris

Re: PATCH: gcc3.0.2 workaround for 8139too

[Luben Tuikov]

gcc 3.0.0, 3.0.1, 3.0.2 needs this patch:

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&pr=4529&database=gcc

to compile the kernel and the 8139too.c nic driver.

-lt




=====
--

__________________________________________________
Do You Yahoo!?
Yahoo! GeoCities - quick and easy web site hosting, just $8.95/month.
http://geocities.yahoo.com/ps/info1

Re: PATCH: gcc3.0.2 workaround for 8139too

[Luben Tuikov]

Sorry, the patch is here...

 http://gcc.gnu.org/ml/gcc-patches/2001-11/msg00698.html

or as someone pointed out get the latest CVS tree.

-lt

--- Luben Tuikov <ltuikov@yahoo.com> wrote:
> gcc 3.0.0, 3.0.1, 3.0.2 needs this patch:
> 
>
http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&pr=4529&database=gcc
> 
> to compile the kernel and the 8139too.c nic driver.
> 
> -lt


=====
--

__________________________________________________
Do You Yahoo!?
Yahoo! GeoCities - quick and easy web site hosting, just $8.95/month.
http://geocities.yahoo.com/ps/info1

Re: PATCH: gcc3.0.2 workaround for 8139too

[Ali Akcaagac]

On Mon, 2001-11-26 at 15:45, Luben Tuikov wrote:
> Sorry, the patch is here...
> 
>  http://gcc.gnu.org/ml/gcc-patches/2001-11/msg00698.html

ahh cool, this verifyies my theory of borked assembly code. well did'nt
paid attention to that patch nor did i knew that it exists.

> or as someone pointed out get the latest CVS tree.

well i use CVS for some stuff but not for CORE things like glibc or gcc.
i've use my 8139too tweak so long that it does'nt' really matters if the
new official gcc 3.0.3 will take a while longer or not.

thank you very much for pointing me to that page.

-- 
Name....: Ali Akcaagac
Status..: Student Of Computer & Economic Science
E-Mail..: mailto:ali.akcaagac@stud.fh-wilhelmshaven.de
WWW.....: http://www.fh-wilhelmshaven.de/~akcaagaa