Re: [alpha, hppa] GCC-4.3 as the default compilers for lenny?

Re: [alpha, hppa] GCC-4.3 as the default compilers for lenny?

[Carlos O'Donell]

On Sat, Mar 22, 2008 at 5:10 PM, Matthias Klose <doko@cs.tu-berlin.de> wrote:
> Carlos O'Donell writes:
>  > >   - gij/gcj shows bus errors on hppa (either 4.2 or 4.3).
>  >
>  > Has gij/gcj ever worked on hppa-linux?
>
>  at least the gij/gcj before adding support for generics (1.5) did
>  work.  Now that a working runtime is required for the compiler makes
>  things different. Please try to run a random HelloWorld program using
>  the interpreter.

I'm not at all interested in gij/gcj. Why do we require that all
languages be functional for a gcc release? This is a difficult goal to
meet.

As the hppa-linux libc-ports maintainer I *am* interested in seeing
the list of regressions you have and commenting on them.

Also note that the hppa-linux list is "linux-parisc@vger.kernel.org"

Cheers,
Carlos.

Re: [alpha, hppa] GCC-4.3 as the default compilers for lenny?

[Matthias Klose]

Carlos O'Donell writes:
> On Sat, Mar 22, 2008 at 5:10 PM, Matthias Klose <doko@cs.tu-berlin.de> wrote:
> > Carlos O'Donell writes:
> >  > >   - gij/gcj shows bus errors on hppa (either 4.2 or 4.3).
> >  >
> >  > Has gij/gcj ever worked on hppa-linux?
> >
> >  at least the gij/gcj before adding support for generics (1.5) did
> >  work.  Now that a working runtime is required for the compiler makes
> >  things different. Please try to run a random HelloWorld program using
> >  the interpreter.
> 
> I'm not at all interested in gij/gcj. Why do we require that all
> languages be functional for a gcc release? This is a difficult goal to
> meet.

I'll let the release team answer this question. IMO we should require
a port to support some kind of java runtime.  Is there any chance to
get the IcedTea Zero port working for hppa, or get the hpux java
support be ported to OpenJDK?

> As the hppa-linux libc-ports maintainer I *am* interested in seeing
> the list of regressions you have and commenting on them.

Please wait for the feedback from Aurelian.

> Also note that the hppa-linux list is "linux-parisc@vger.kernel.org"

thanks for the hint.

  Matthias

Re: [alpha, hppa] GCC-4.3 as the default compilers for lenny?

[Carlos O'Donell]

On Sat, Mar 22, 2008 at 6:10 PM, Aurelien Jarno <aurelien@aurel32.net> wrote:
>  I have tried to build the glibc with GCC 4.3 on hppa, and rpcgen
>  segfaults when it is used, so the build fails. I haven't start to
>  investigate the problem (I started by the architectures where the
>  problems were minor).

I switched from gcc-4_2-branch to gcc-4_3-branch and I see the loader
gets a SIGTRAP, not a SIGSEGV. I'll try to debug this.

Cheers,
Carlos.

Re: [alpha, hppa] GCC-4.3 as the default compilers for lenny?

[Grant Grundler]

On Tue, Mar 25, 2008 at 06:42:44PM -0400, Carlos O'Donell wrote:
> On Sat, Mar 22, 2008 at 6:10 PM, Aurelien Jarno <aurelien@aurel32.net> wrote:
> >  I have tried to build the glibc with GCC 4.3 on hppa, and rpcgen
> >  segfaults when it is used, so the build fails. I haven't start to
> >  investigate the problem (I started by the architectures where the
> >  problems were minor).
> 
> I switched from gcc-4_2-branch to gcc-4_3-branch and I see the loader
> gets a SIGTRAP, not a SIGSEGV. I'll try to debug this.


Another gcc problem report:

That past weekend I built the latest parisc-2.6-25-rc6 kernel from
Kyle's tree using gcc-4.1, gcc-4.2, and gcc-4.3. All three kernels
booted but the networking only worked for gcc-4.1 kernel.

When sending ICMP (ping) messages, the "ifconfig eth0" output on the
parisc machine was reporting RX and TX packets at the expected rate
(1 per second).  But the source host (omnibook 500 laptop) only rarely
saw the ICMP reply. Maybe 1 out of 30 or 50 ICMP.  My guess is the
IP layer is racing with DMA and the packet is mangled in some subtle way.
Both gcc-4.2 and gcc-4.3 showed the same symptom.

I just realized it's possible the 100BT switch is dropping the packets.
I'll try to capture a tcpdump from the omnibook 500 side using a
crossover cable tonight or tomorrow since without that, I don't see
how this can be debugged.

thanks,
grant

Re: [alpha, hppa] GCC-4.3 as the default compilers for lenny?

[Grant Grundler]

On Wed, Mar 26, 2008 at 10:20:35AM -0600, Grant Grundler wrote:
> Another gcc problem report:
> 
> That past weekend I built the latest parisc-2.6-25-rc6 kernel from
> Kyle's tree using gcc-4.1, gcc-4.2, and gcc-4.3. All three kernels
> booted but the networking only worked for gcc-4.1 kernel.

Sorry, I just realized I didn't mention the gcc versions are whatever is
currently in "Debian testing" for parisc.

grant

Re: [alpha, hppa] GCC-4.3 as the default compilers for lenny?

[John David Anglin]

> On Wed, Mar 26, 2008 at 10:20:35AM -0600, Grant Grundler wrote:
> > Another gcc problem report:
> > 
> > That past weekend I built the latest parisc-2.6-25-rc6 kernel from
> > Kyle's tree using gcc-4.1, gcc-4.2, and gcc-4.3. All three kernels
> > booted but the networking only worked for gcc-4.1 kernel.
> 
> Sorry, I just realized I didn't mention the gcc versions are whatever is
> currently in "Debian testing" for parisc.

I would look carefully to see if the newer compilers generate warnings
that need fixing.

I've been using 4.1 for my recent builds.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc-cnrc.gc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6602)

Re: [alpha, hppa] GCC-4.3 as the default compilers for lenny?

[Grant Grundler]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=unknown-8bit, Size: 2875 bytes --]

On Wed, Mar 26, 2008 at 10:20:35AM -0600, Grant Grundler wrote:
> Another gcc problem report:
>=20
> That past weekend I built the latest parisc-2.6-25-rc6 kernel from
> Kyle's tree using gcc-4.1, gcc-4.2, and gcc-4.3. All three kernels
> booted but the networking only worked for gcc-4.1 kernel.

=46ollowup: gcc-42 is working too now and I'm not sure why.
I upgraded gcc-4.3 package from 4.3.0-rcX to 4.3.1 AND all of it's
dependencies as well.

I looked at gcc-43 build output to check warnings related to networking
as suggest by John David Anglin.

I ignored the couple of warnings in drivers/net/wireless/.

  CC [M]  drivers/net/fealnx.o
drivers/net/fealnx.c:896:2: warning: #warning Processor architecture un=
defined!
    No clue what this driver is for. Ignored.

  CC [M]  drivers/net/lasi_82596.o
drivers/net/lasi_82596.c: In function =E2=80=98lan_init_chip=E2=80=99:
drivers/net/lasi_82596.c:164: warning: format =E2=80=98%lx=E2=80=99 exp=
ects type =E2=80=98long unsigned int=E2=80=99, but argument 3 has type =
=E2=80=98resource_size_t=E2=80=99
drivers/net/lasi_82596.c:169: warning: format =E2=80=98%lx=E2=80=99 exp=
ects type =E2=80=98long unsigned int=E2=80=99, but argument 2 has type =
=E2=80=98resource_size_t=E2=80=99

	That's just printk format strings.

overshot and found this one in drivers/scsi/ interesting too:

drivers/scsi/initio.c:2585: warning: large integer implicitly truncated=
 to unsigned type

matching code is:
        cblk->senselen =3D cpu_to_le32(SENSE_SIZE);

grundler@j6k:~/parisc-2.6-gcc43$ fgrep senselen drivers/scsi/*.h
drivers/scsi/aha1740.h: u8 senselen;            /* Sense Length */
drivers/scsi/initio.h:  u8 senselen;    /*31 Sense Allocation Length */
drivers/scsi/megaraid.h:        u8 reqsenselen;

The code at first glance looks wrong here. Patch submitted to linux-scs=
i
and Alan Cox (who appears to have taken on this driver last year).


Nothing else in the drivers/net.
A bit more in net/ though:
  CC      net/sunrpc/svc.o
net/sunrpc/svc.c: In function =E2=80=98__svc_create_thread=E2=80=99:
net/sunrpc/svc.c:587: warning: =E2=80=98oldmask.bits[0u]=E2=80=99 may b=
e used uninitialized in this function

	Don't care since I'm not using rpc.
	This is probably a false positive anyway. I've found several.

And that was it. So no warnings that seem to be related to this network=
ing problem. It was worth checking though.


> I just realized it's possible the 100BT switch is dropping the packet=
s.
> I'll try to capture a tcpdump from the omnibook 500 side using a
> crossover cable tonight or tomorrow since without that, I don't see
> how this can be debugged.

Still need to try the cross-over cable.

grant
--
To unsubscribe from this list: send the line "unsubscribe linux-parisc"=
 in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [alpha, hppa] GCC-4.3 as the default compilers for lenny?

[rubisher]

Hello Grant,

Grant Grundler wrote:
> On Wed, Mar 26, 2008 at 10:20:35AM -0600, Grant Grundler wrote:
>> Another gcc problem report:
>>
>> That past weekend I built the latest parisc-2.6-25-rc6 kernel from
>> Kyle's tree using gcc-4.1, gcc-4.2, and gcc-4.3. All three kernels
>> booted but the networking only worked for gcc-4.1 kernel.
>=20
> Followup: gcc-42 is working too now and I'm not sure why.
> I upgraded gcc-4.3 package from 4.3.0-rcX to 4.3.1 AND all of it's
> dependencies as well.
>=20
> I looked at gcc-43 build output to check warnings related to networki=
ng
> as suggest by John David Anglin.
>=20
> I ignored the couple of warnings in drivers/net/wireless/.
>=20
>   CC [M]  drivers/net/fealnx.o
> drivers/net/fealnx.c:896:2: warning: #warning Processor architecture =
undefined!
>     No clue what this driver is for. Ignored.
>=20
>   CC [M]  drivers/net/lasi_82596.o
> drivers/net/lasi_82596.c: In function =91lan_init_chip=92:
> drivers/net/lasi_82596.c:164: warning: format =91%lx=92 expects type =
=91long unsigned int=92, but argument 3 has type =91resource_size_t=92
> drivers/net/lasi_82596.c:169: warning: format =91%lx=92 expects type =
=91long unsigned int=92, but argument 2 has type =91resource_size_t=92
>=20
> 	That's just printk format strings.
>=20
That remember me that when I discover those new warnings with gcc-4.2 <=
http://article.gmane.org/gmane.linux.ports.parisc/70>
"discards qualifiers from pointer target type", they disapeared with gc=
c-4.3 with and btw this nic wasn't working anymore=20
(no ping no ssh, ...) even thought the kernel was booting. Unfortunatel=
y, I didn't have time to investigate in more details=20
(and it wasn't question to use gcc-4.3 as new default), sorry. If you c=
ompare your kernel build output with gcc-4.2 and 4.3=20
do you also notice this differences? (even thought it could be not rela=
ted to the issue I encountered, though).

Tx,
	r.


> overshot and found this one in drivers/scsi/ interesting too:
>=20
> drivers/scsi/initio.c:2585: warning: large integer implicitly truncat=
ed to unsigned type
>=20
> matching code is:
>         cblk->senselen =3D cpu_to_le32(SENSE_SIZE);
>=20
> grundler@j6k:~/parisc-2.6-gcc43$ fgrep senselen drivers/scsi/*.h
> drivers/scsi/aha1740.h: u8 senselen;            /* Sense Length */
> drivers/scsi/initio.h:  u8 senselen;    /*31 Sense Allocation Length =
*/
> drivers/scsi/megaraid.h:        u8 reqsenselen;
>=20
> The code at first glance looks wrong here. Patch submitted to linux-s=
csi
> and Alan Cox (who appears to have taken on this driver last year).
>=20
>=20
> Nothing else in the drivers/net.
> A bit more in net/ though:
>   CC      net/sunrpc/svc.o
> net/sunrpc/svc.c: In function =91__svc_create_thread=92:
> net/sunrpc/svc.c:587: warning: =91oldmask.bits[0u]=92 may be used uni=
nitialized in this function
>=20
> 	Don't care since I'm not using rpc.
> 	This is probably a false positive anyway. I've found several.
>=20
> And that was it. So no warnings that seem to be related to this netwo=
rking problem. It was worth checking though.
>=20
>=20
>> I just realized it's possible the 100BT switch is dropping the packe=
ts.
>> I'll try to capture a tcpdump from the omnibook 500 side using a
>> crossover cable tonight or tomorrow since without that, I don't see
>> how this can be debugged.
>=20
> Still need to try the cross-over cable.
>=20
> grant

Hello *,

Some time ago
--
To unsubscribe from this list: send the line "unsubscribe linux-parisc"=
 in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [alpha, hppa] GCC-4.3 as the default compilers for lenny?

[Grant Grundler]

On Sat, Mar 29, 2008 at 06:24:12PM +0000, rubisher wrote:
...
>> I looked at gcc-43 build output to check warnings related to networking
>> as suggest by John David Anglin.
...
> That remember me that when I discover those new warnings with gcc-4.2 
> <http://article.gmane.org/gmane.linux.ports.parisc/70>
> "discards qualifiers from pointer target type", they disapeared with 
> gcc-4.3

I saw that but don't know what it's about.

> and btw this nic wasn't working anymore (no ping no ssh, ...) 
> even thought the kernel was booting. Unfortunately, I didn't have time to 
> investigate in more details (and it wasn't question to use gcc-4.3 as new 
> default), sorry. If you compare your kernel build output with gcc-4.2 and 
> 4.3 do you also notice this differences?

By "build output" do you mean the binary (e.g. dot-o files) or log of what
was compiled and which warnings were emitted?

thanks
grant

Re: [alpha, hppa] GCC-4.3 as the default compilers for lenny?

[rubisher]

Grant Grundler wrote:
> On Sat, Mar 29, 2008 at 06:24:12PM +0000, rubisher wrote:
> ...
>>> I looked at gcc-43 build output to check warnings related to networking
>>> as suggest by John David Anglin.
> ...
>> That remember me that when I discover those new warnings with gcc-4.2 
>> <http://article.gmane.org/gmane.linux.ports.parisc/70>
>> "discards qualifiers from pointer target type", they disapeared with 
>> gcc-4.3
> 
> I saw that but don't know what it's about.
> 
The answer of Matthew seems to work ;-)

>> and btw this nic wasn't working anymore (no ping no ssh, ...) 
>> even thought the kernel was booting. Unfortunately, I didn't have time to 
>> investigate in more details (and it wasn't question to use gcc-4.3 as new 
>> default), sorry. If you compare your kernel build output with gcc-4.2 and 
>> 4.3 do you also notice this differences?
> 
> By "build output" do you mean the binary (e.g. dot-o files) or log of what
> was compiled and which warnings were emitted?
> 
Well, my understanding was that you compared a log file of a build with gcc-4.2 with the log file of the build of same src 
with gcc-4.3. That's what I did in fact but I was so busy this week that I couldn't post the results. Sorry, I will be off 
next week too, so will try to do when I will be back next Apr 14 ;-)

hth,
	r.

> thanks
> grant
> 
>