Re: [PATCH] Replace --enable-static-pie with --disable-default-pie

[Siddhesh Poyarekar <siddhesh@sourceware.org>]

On 12/7/21 21:36, Florian Weimer wrote:
> * Siddhesh Poyarekar:
> 
>> Build glibc programs and tests as PIE by default and enable static-pie
>> automatically if the architecture and toolchain supports it.
>>
>> Also add a new configuration option --disable-default-pie to prevent
>> building programs as PIE.
>>
>> Only the following architectures now have PIE disabled by default
>> because they do not work at the moment.  hppa, ia64 and csky don't work
>> because the linker is unable to handle a pcrel relocation generated from
>> PIE objects.  The microblaze compiler is currently failing with an ICE.
>> GNU hurd tries to enable static-pie, which does not work and hence
>> fails.  It could be made to work, but I've left the enabling for the
>> Hurd maintainer.  build-many-glibcs runs clean for all targets now;
>> following are the failures before disabling PIE for these targets.  I
>> also tested x86_64 on Fedora and Ubuntu, to verify that the default
>> build as well as --disable-default-pie work as expected with both system
>> toolchains.
> 
> I see a “make check” failure on alpha:
> 
> /home/bmg/build/glibcs/alpha-linux-gnu/glibc/math/test-misc.o: in function `do_test':
> /home/bmg/src/glibc/math/test-misc.c:391:(.text+0x780): relocation truncated to fit: GPREL16 against `.rodata.cst16'
> /home/bmg/src/glibc/math/test-misc.c:400:(.text+0x87c): relocation truncated to fit: GPREL16 against `.rodata.cst16'
> /home/bmg/src/glibc/math/test-misc.c:705:(.text+0xda0): relocation truncated to fit: GPREL16 against `.rodata.cst16'
> /home/bmg/src/glibc/math/test-misc.c:714:(.text+0xe90): relocation truncated to fit: GPREL16 against `.rodata.cst16'
> collect2: error: ld returned 1 exit status
> make[3]: *** [../Rules:230: /home/bmg/build/glibcs/alpha-linux-gnu/glibc/math/test-misc] Error 1
> 
> This is with default build-many-glibcs.py.

My alpha `compilers` build had failed despite manually rebasing the 
binutils and gcc release branches.  That's similar to the hppa failure 
where the linker isn't able to handle a relative relocation.  I'll 
disable PIE for alpha too.

> The NEWS entry should mention that --enable-static-pie is now ignored
> (and no longer necessary).

I did mention it...

>  Deprecated and removed features, and other changes affecting compatibility:
>  
>  * The r_version update in the debugger interface makes the glibc binary
> @@ -80,6 +88,10 @@ Deprecated and removed features, and other changes affecting compatibility:
>  
>  * Intel MPX support (lazy PLT, ld.so profile, and LD_AUDIT) has been removed.
>  
> +* The --enable-static-pie option is no longer available.  The glibc build
> +  configuration script now automatically detects static-pie support in the
> +  toolchain and architecture and enables it if available.
> +
>  Changes to build and runtime requirements:
>  
>    [Add changes to build and runtime requirements here]

... but I just noticed the "Changes to build and runtime requirements" 
section.  Should I just consolidate the NEWS note into this section instead?

Thanks,
Siddhesh