DMARC-Filter: OpenDMARC Filter v1.3.2 mx1.redhat.com 218CA4E33A
Date: Mon, 25 Sep 2017 16:03:13 +0800
From: Dave Young <dyoung@redhat.com>
To: AKASHI Takahiro <takahiro.akashi@linaro.org>, catalin.marinas@arm.com,
        will.deacon@arm.com, bauerman@linux.vnet.ibm.com, dhowells@redhat.com,
        vgoyal@redhat.com, herbert@gondor.apana.org.au, davem@davemloft.net,
        akpm@linux-foundation.org, mpe@ellerman.id.au, bhe@redhat.com,
        arnd@arndb.de, ard.biesheuvel@linaro.org, kexec@lists.infradead.org,
        linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org
Subject: Re: [PATCH v3 03/10] kexec_file: factor out arch_kexec_kernel_*()
 from x86, powerpc
Message-ID: <20170925080313.GA5970@dhcp-128-65.nay.redhat.com>
References: <20170915105932.25338-1-takahiro.akashi@linaro.org>
 <20170915105932.25338-4-takahiro.akashi@linaro.org>
 <20170921073516.GA6643@dhcp-128-65.nay.redhat.com>
 <20170922075807.GM17186@linaro.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20170922075807.GM17186@linaro.org>
User-Agent: Mutt/1.8.3 (2017-05-23)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 3863
Lines: 107

HI AKASHI,
On 09/22/17 at 04:58pm, AKASHI Takahiro wrote:
> Hi Dave,
> 
[snip]

> > >  /*
> > >   * Apply purgatory relocations.
> > >   *
> > > diff --git a/include/linux/kexec.h b/include/linux/kexec.h
> > > index dd056fab9e35..4a2b24d94e04 100644
> > > --- a/include/linux/kexec.h
> > > +++ b/include/linux/kexec.h
> > > @@ -134,6 +134,26 @@ struct kexec_file_ops {
> > >  #endif
> > >  };
> > >  
> > > +int kexec_kernel_image_probe(struct kimage *image, void *buf,
> > > +			     unsigned long buf_len);
> > > +void *kexec_kernel_image_load(struct kimage *image);
> > > +int kexec_kernel_post_load_cleanup(struct kimage *image);
> > > +int kexec_kernel_verify_sig(struct kimage *image, void *buf,
> > > +			    unsigned long buf_len);
> > > +
> > > +#ifndef arch_kexec_kernel_image_probe
> > > +#define arch_kexec_kernel_image_probe kexec_kernel_image_probe
> > > +#endif
> > > +#ifndef arch_kexec_kernel_image_load
> > > +#define arch_kexec_kernel_image_load kexec_kernel_image_load
> > > +#endif
> > > +#ifndef arch_kimage_file_post_load_cleanup
> > > +#define arch_kimage_file_post_load_cleanup kexec_kernel_post_load_cleanup
> > > +#endif
> > > +#ifndef arch_kexec_kernel_verify_sig
> > > +#define arch_kexec_kernel_verify_sig kexec_kernel_verify_sig
> > > +#endif
> > > +
> > >  /**
> > >   * struct kexec_buf - parameters for finding a place for a buffer in memory
> > >   * @image:	kexec image in which memory to search.
> > > @@ -276,12 +296,6 @@ int crash_shrink_memory(unsigned long new_size);
> > >  size_t crash_get_memory_size(void);
> > >  void crash_free_reserved_phys_range(unsigned long begin, unsigned long end);
> > >  
> > > -int __weak arch_kexec_kernel_image_probe(struct kimage *image, void *buf,
> > > -					 unsigned long buf_len);
> > > -void * __weak arch_kexec_kernel_image_load(struct kimage *image);
> > > -int __weak arch_kimage_file_post_load_cleanup(struct kimage *image);
> > > -int __weak arch_kexec_kernel_verify_sig(struct kimage *image, void *buf,
> > > -					unsigned long buf_len);
> > 
> > I thought we can keep using the __weak function in common code and drop
> > the arch functions, no need the #ifndef arch_kexec_kernel_image_probe
> > and the function renaming stuff.  But I did not notice the powerpc
> > _probe function checks KEXEC_ON_CRASH, that should be checked earlier
> > and we can fail out early if not supported, but I have no idea
> > how to do it gracefully for now.
> > 
> > Also for x86 _load function it cleanups image->arch.elf_headers, it can
> > not be dropped simply.
> 
> Yeah, arm64 post_load_cleanup function also has some extra stuff.
> See my patch #7/8.

But the x86 cleanup was dropped silently, can you add it in x86
post_load_cleanup as well?

> 
> > Consider the above two issues, maybe you can keep the powerpc
> > version of _probe() and x86 version of _load(), and still copy the common code
> > to kexec_file.c weak functions.
> 
> It was exactly what I made before submitting v3, but I changed
> my mind a bit. My intension is to prevent the code being duplicated
> even though it has only a few lines of code.
> 
> I agree that '#ifndef arch_kexec_kernel_image_probe' in kexec.h would be
> quite ugly, but similar usages can be spotted in the kernel source.
> 
> That said if you don't like it at all, I defer to you.

I understand your concern, maybe still use a weak function for 
arch_kexec_kernel_image_*, and they call the kexec_kernel_image_* in
kexec_file.c common code.

Like in general code:

int __weak arch_kexec_kernel_image_probe(struct kimage *image, void *buf,
                                         unsigned long buf_len)
{
	return kexec_kernel_image_probe(image, buf, buf_len);
}

In architechture code it can add other code and call
kexec_kernel_image_*

It looks a bit better than the #ifdef way. 

[snip]

> 
> Thanks,
> -Takahiro AKASHI
> 

Thanks
Dave