Re: problem compiling kernel ld error relocation truncated to fit: GPREL16 against symbol

Re: problem compiling kernel ld error relocation truncated to fit: GPREL16 against symbol

[Michael Cree]

[Added linux-alpha to CC to increase likelihood of reaching someone who
best knows about this issue.]

On 03/11/11 15:54, carl hansen wrote:
> Problem compiling kernel 3.1, gcc 4.6.1 or .2, latest binutils
> 
> I think it's ld error
> google suggested -Wl,--no-relax, didn't make a difference, unless I
> did  it wrong

The binutils bug that can be worked around with --no-relax is not the
issue here.

>  ld  -static -N  --build-id -o .tmp_vmlinux1 -T
> arch/alpha/kernel/vmlinux.lds arch/alpha/kernel/head.o
> init/built-in.o --start-group  usr/built-in.o
> arch/alpha/kernel/built-in.o  arch/alpha/mm/built-in.o
> arch/alpha/math-emu/built-in.o  kernel/built-in.o  mm/built-in.o
> fs/built-in.o  ipc/built-in.o  security/built-in.o  crypto/built-in.o
> block/built-in.o  lib/lib.a  arch/alpha/lib/lib.a  lib/built-in.o
> arch/alpha/lib/built-in.o  drivers/built-in.o  sound/built-in.o
> firmware/built-in.o  net/built-in.o --end-group
> init/built-in.o: In function `do_one_initcall':
> /var/src.build/kernelbuild/latest/linux-3.1/init/main.c:682:(.text+0x8):
> relocation truncated to fit: GPREL16 against symbol `initcall_debug'
> defined in .sbss section in init/built-in.o

Confirmed.  Building kernel 3.1 with config for Miata and mainly only
drivers needed for that hardware compiles fine with gcc-4.4 but fails
with the GPREL16 link errors when compiled with gcc-4.6.

If I modify arch/kernel/Makefile to add -mlarge-data rather than
-msmall-data to CFLAGS then the kernel successfully compiles with gcc-4.6.

Is there anyone more knowledgeable than I about gcc who can give us
guidance as to why gcc-4.6 should be enlarging the data segment so much
that the small data model cannot be used?  Or is there a fix required in
the kernel code?

Cheers
Michael.

Re: problem compiling kernel ld error relocation truncated to fit: GPREL16 against symbol

[Raúl Porcel]

On 11/13/2011 01:43 AM, Michael Cree wrote:
[snip]
> 
> Is there anyone more knowledgeable than I about gcc who can give us
> guidance as to why gcc-4.6 should be enlarging the data segment so much
> that the small data model cannot be used?  Or is there a fix required in
> the kernel code?

Guess Uros can help here...

Re: problem compiling kernel ld error relocation truncated to fit: GPREL16 against symbol

[Uros Bizjak]

On Tue, Dec 6, 2011 at 5:46 PM, Raúl Porcel <armin76@gentoo.org> wrote:

> On 11/13/2011 01:43 AM, Michael Cree wrote:
> [snip]
>>
>> Is there anyone more knowledgeable than I about gcc who can give us
>> guidance as to why gcc-4.6 should be enlarging the data segment so much
>> that the small data model cannot be used?  Or is there a fix required in
>> the kernel code?
>
> Guess Uros can help here...

Unfortunately, I have no idea... There are similar problems with gcc
build, so I use:

# Prevent GPREL16 relocation truncation
LDFLAGS += -Wl,--no-relax
BOOT_LDFLAGS += -Wl,--no-relax

to finish the bootstrap. IMO, this points to linker problems.

Perhaps rth (CC'd) can say something about this problem.

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

Re: problem compiling kernel ld error relocation truncated to fit: GPREL16 against symbol

[Michael Cree]

On 7/12/2011, at 7:55 AM, Uros Bizjak wrote:
> On Tue, Dec 6, 2011 at 5:46 PM, Raúl Porcel <armin76@gentoo.org>  
> wrote:
>> On 11/13/2011 01:43 AM, Michael Cree wrote:
>> [snip]
>>>
>>> Is there anyone more knowledgeable than I about gcc who can give us
>>> guidance as to why gcc-4.6 should be enlarging the data segment so  
>>> much
>>> that the small data model cannot be used?  Or is there a fix  
>>> required in
>>> the kernel code?
>>
>> Guess Uros can help here...
>
> Unfortunately, I have no idea... There are similar problems with gcc
> build, so I use:
>
> # Prevent GPREL16 relocation truncation
> LDFLAGS += -Wl,--no-relax
> BOOT_LDFLAGS += -Wl,--no-relax

Yeah, there are a growing number of packages being affected by this  
(presumably linker) problem and I have recently seen it with gcc- 
snapshot and gnat-4.6 in Debian.  However I don't think this is the  
issue seen with the Linux kernel, which has the --relax argument  
commented out of the linker flags in the Makefile.

What we see is that gcc-4.4 successfully compiles the Linux kernel.   
Using gcc-4.6 results in relocation errors.  The kernel Makefile adds - 
msmall-data to the compiler flags.  If this is change to -mlarge-data  
then compilation with gcc-4.6 succeeds.  What I am wondering about is  
what is the change in gcc-4.6 that causes the data segment  
(presumably) to be too large to stored in 64 kbytes (whereas gcc-4.4  
can fit it in 64kbytes)?

Cheers
Michael.

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