Date:	Tue, 31 May 2016 22:58:59 +0200
From:	Ard Biesheuvel <ard.biesheuvel@...aro.org>
To:	Dmitry Shmidt <dimitrysh@...gle.com>
Cc:	Arend Van Spriel <arend.vanspriel@...adcom.com>,
	Russell King <linux@....linux.org.uk>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"<linux-arm-kernel@...ts.infradead.org>" 
	<linux-arm-kernel@...ts.infradead.org>,
	Hante Meuleman <hante.meuleman@...adcom.com>
Subject: Re: building and using modules on arm64 hikey board

On 31 May 2016 at 22:24, Dmitry Shmidt <dimitrysh@...gle.com> wrote:
> On Mon, May 30, 2016 at 4:30 AM, Ard Biesheuvel
> <ard.biesheuvel@...aro.org> wrote:
>> This is likely caused by the fact that the Android AArch64 toolchain uses -fpic by default. Could you try adding -fno-pic to the CFLAGS?
>
> Actually Arend is using 4.4, and we need to pull your fix, Ard:
>
> commit 80a2d83376001f6a1993f2e925670ab0e4cdb76d
> Author: Ard Biesheuvel <ard.biesheuvel@...aro.org>
> Date:   Tue Jan 5 10:18:52 2016 +0100
>
>     arm64: module: avoid undefined shift behavior in reloc_data()
>

OK, that was going to be my next question to Arend, i.e., to check
whether he has all the recent fixes we did for the module loader.

But I'd also like to understand how we ended up with PREL32
relocations in the first place, since those are quite unusual in
object code generated from generic C source code when using the
non-pic small model, which is normally GCC's default. Are there any
other code generation defaults for the Android  AArch64 GCC toolchain
that you are aware of?

>>> On 30 mei 2016, at 12:21, Arend Van Spriel <arend.vanspriel@...adcom.com> wrote:
>>>
>>> I got myself an arm64 HiKey board from LeMaker and build an Android AOSP
>>> image for it (see [1]). For development I would like to use
>>> CONFIG_MODULES. However, when I try to insmod the build module I get:
>>>
>>> [  287.903653] module cfg80211: overflow in relocation type 261 val
>>> ffffffbffc006530
>>>
>>> Looking AArch64 ELF documentation [2], section 4.6.5, it has:
>>> code|name            |operation |overflow check   |
>>> 261 |R_AARCH64_PREL32|S + A - P |-2^31 <= X < 2^32|
>>>
>>> So basically the highest 32 bits should all be one and so ffffffbf is
>>> invalid. From what I could find searching internet it could be an issue
>>> with linker options so I build kernel and modules with V=1. Here the
>>> linker invocation for them:
>>>
>>> + aarch64-linux-android-ld -EL -p --no-undefined -X --build-id -o vmlinux \
>>> -T ./arch/arm64/kernel/vmlinux.lds arch/arm64/kernel/head.o
>>> init/built-in.o \
>>> --start-group usr/built-in.o arch/arm64/kernel/built-in.o
>>> arch/arm64/mm/built-in.o \
>>> arch/arm64/net/built-in.o arch/arm64/kvm/built-in.o
>>> arch/arm64/crypto/built-in.o \
>>> ./drivers/firmware/efi/libstub/lib.a kernel/built-in.o certs/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 \
>>> arch/arm64/lib/lib.a lib/lib.a arch/arm64/lib/built-in.o lib/built-in.o
>>> drivers/built-in.o \
>>> sound/built-in.o firmware/built-in.o net/built-in.o virt/built-in.o
>>> --end-group .tmp_kallsyms2.o
>>>
>>>  aarch64-linux-android-ld -EL -r  -T ./scripts/module-common.lds
>>> --build-id \
>>>  -o net/wireless/cfg80211.ko net/wireless/cfg80211.o
>>> net/wireless/cfg80211.mod.o
>>>
>>> Attached are vmlinux.lds and module-common.lds. I also tried taking
>>> upstream arch/arm64/kernel/module.lds in hikey-linaro tree. If someone
>>> can give a hint or educated guess at what to try it would be appreciated.
>>>
>>> Regards,
>>> Arend
>>>
>>> [1] https://source.android.com/source/devices.html#building-kernel
>>> [2]
>>> http://infocenter.arm.com/help/topic/com.arm.doc.ihi0056b/IHI0056B_aaelf64.pdf
>>> <module-common.lds>
>>> <vmlinux.lds>
>>
>> _______________________________________________
>> linux-arm-kernel mailing list
>> linux-arm-kernel@...ts.infradead.org
>> http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

Powered by blists - more mailing lists