Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Wed, 11 Sep 2019 17:40:13 +0100
From:   Will Deacon <will@kernel.org>
To:     Jessica Yu <jeyu@kernel.org>
Cc:     linux-kernel@vger.kernel.org, maco@android.com,
        gregkh@linuxfoundation.org,
        Matthias Maennich <maennich@google.com>,
        Ard Biesheuvel <ard.biesheuvel@linaro.org>,
        Catalin Marinas <catalin.marinas@arm.com>
Subject: Re: [PATCH] module: Fix link failure due to invalid relocation on
 namespace offset
Message-ID: <20190911164012.nalsccw6jku7gbpw@willie-the-truck>
References: <20190911122646.13838-1-will@kernel.org>
 <20190911133506.GB7837@linux-8ccs>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20190911133506.GB7837@linux-8ccs>
User-Agent: NeoMutt/20170113 (1.7.2)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, Sep 11, 2019 at 03:35:06PM +0200, Jessica Yu wrote:
> +++ Will Deacon [11/09/19 13:26 +0100]:
> > Commit 8651ec01daed ("module: add support for symbol namespaces.")
> > broke linking for arm64 defconfig:
> > 
> >  | lib/crypto/arc4.o: In function `__ksymtab_arc4_setkey':
> >  | arc4.c:(___ksymtab+arc4_setkey+0x8): undefined reference to `no symbol'
> >  | lib/crypto/arc4.o: In function `__ksymtab_arc4_crypt':
> >  | arc4.c:(___ksymtab+arc4_crypt+0x8): undefined reference to `no symbol'
> > 
> > This is because the dummy initialisation of the 'namespace_offset' field
> > in 'struct kernel_symbol' when using EXPORT_SYMBOL on architectures with
> > support for PREL32 locations uses an offset from an absolute address (0)
> > in an effort to trick 'offset_to_pointer' into behaving as a NOP,
> > allowing non-namespaced symbols to be treated in the same way as those
> > belonging to a namespace.
> > 
> > Unfortunately, place-relative relocations require a symbol reference
> > rather than an absolute value and, although x86 appears to get away with
> > this due to placing the kernel text at the top of the address space, it
> > almost certainly results in a runtime failure if the kernel is relocated
> > dynamically as a result of KASLR.
> > 
> > Rework 'namespace_offset' so that a value of 0, which cannot occur for a
> > valid namespaced symbol, indicates that the corresponding symbol does
> > not belong to a namespace.
> > 
> > Cc: Matthias Maennich <maennich@google.com>
> > Cc: Jessica Yu <jeyu@kernel.org>
> > Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
> > Cc: Catalin Marinas <catalin.marinas@arm.com>
> > Fixes: 8651ec01daed ("module: add support for symbol namespaces.")
> > Reported-by: kbuild test robot <lkp@intel.com>
> > Signed-off-by: Will Deacon <will@kernel.org>
> > ---
> > 
> > Please note that I've not been able to test this at LPC, but it's been
> > submitted to kernelci.
> 
> Thanks for fixing this so quickly. I can confirm that this fixes the
> build for arm64 defconfig and x86 built fine for me as well. I'll wait
> a bit and apply this at the end of the day in case Matthias or anybody
> else would like to confirm/test.

FWIW, I've managed to boot arm64 Debian under QEMU and load/unload
modules successfully with this patch applied on top of modules-next.

Will