[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <43ef1623-2186-aa07-cfa5-d618aa4f09c0@ghiti.fr>
Date: Thu, 11 Jun 2020 15:43:56 -0400
From: Alex Ghiti <alex@...ti.fr>
To: Jerome Forissier <jerome@...issier.org>,
Michael Ellerman <mpe@...erman.id.au>,
Benjamin Herrenschmidt <benh@...nel.crashing.org>,
Paul Mackerras <paulus@...ba.org>,
Paul Walmsley <paul.walmsley@...ive.com>,
Palmer Dabbelt <palmer@...belt.com>,
Albert Ou <aou@...s.berkeley.edu>,
Anup Patel <Anup.Patel@....com>,
Atish Patra <Atish.Patra@....com>,
Zong Li <zong.li@...ive.com>, linux-kernel@...r.kernel.org,
linuxppc-dev@...ts.ozlabs.org, linux-riscv@...ts.infradead.org
Cc: Anup Patel <anup@...infault.org>
Subject: Re: [PATCH v5 2/4] riscv: Introduce CONFIG_RELOCATABLE
Hi Jerome,
Le 6/10/20 à 10:10 AM, Jerome Forissier a écrit :
> On 6/7/20 9:59 AM, Alexandre Ghiti wrote:
> [...]
>
>> +config RELOCATABLE
>> + bool
>> + depends on MMU
>> + help
>> + This builds a kernel as a Position Independent Executable (PIE),
>> + which retains all relocation metadata required to relocate the
>> + kernel binary at runtime to a different virtual address than the
>> + address it was linked at.
>> + Since RISCV uses the RELA relocation format, this requires a
>> + relocation pass at runtime even if the kernel is loaded at the
>> + same address it was linked at.
> Is this true? I thought that the GNU linker would write the "proper"
> values by default, contrary to the LLVM linker (ld.lld) which would need
> a special flag: --apply-dynamic-relocs (by default the relocated places
> are set to zero). At least, it is my experience with Aarch64 on a
> different project. So, sorry if I'm talking nonsense here -- I have not
> looked at the details.
>
>
It seems that you're right, at least for aarch64 since they specifically
specify the --no-apply-dynamic-relocs option. I retried to boot without
relocating at runtime, and it fails on riscv. Can this be arch specific ?
Thanks,
Alex
Powered by blists - more mailing lists