From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751973AbdJXJi2 (ORCPT ); Tue, 24 Oct 2017 05:38:28 -0400 Received: from pandora.armlinux.org.uk ([78.32.30.218]:57168 "EHLO pandora.armlinux.org.uk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932308AbdJXJiZ (ORCPT ); Tue, 24 Oct 2017 05:38:25 -0400 Date: Tue, 24 Oct 2017 10:38:15 +0100 From: Russell King - ARM Linux To: Ard Biesheuvel Cc: Chris Zhong , jeffy , "linux-kernel@vger.kernel.org" , "linux-arm-kernel@lists.infradead.org" , Ingo Molnar Subject: Re: [PATCH] ARM: Fix zImage file size not aligned with CONFIG_EFI_STUB enabled Message-ID: <20171024093814.GB20805@n2100.armlinux.org.uk> References: <20171023085006.GM20805@n2100.armlinux.org.uk> <59EDC34C.40109@rock-chips.com> <20171023105046.GT20805@n2100.armlinux.org.uk> <20171023114549.GU20805@n2100.armlinux.org.uk> <20171024090953.GZ20805@n2100.armlinux.org.uk> <20171024092229.GA20805@n2100.armlinux.org.uk> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.23 (2014-03-12) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Tue, Oct 24, 2017 at 10:30:41AM +0100, Ard Biesheuvel wrote: > On 24 October 2017 at 10:26, Ard Biesheuvel wrote: > > On 24 October 2017 at 10:22, Russell King - ARM Linux > > wrote: > >> On Tue, Oct 24, 2017 at 10:13:09AM +0100, Ard Biesheuvel wrote: > >>> On 24 October 2017 at 10:09, Russell King - ARM Linux > >>> wrote: > >>> > The question is: do we want to know when additional sections get > >>> > emitted into the binary? > >>> > >>> Well, we need to know whether the size of zImage is a multiple of 512 > >>> bytes. We could check that separately by adding the following as well > >>> > >>> #ifdef CONFIG_EFI_STUB > >>> ASSERT((_edata % 512 == 0), "zImage file size is not a multiple of 512 bytes") > >> > >> ASSERT((_edata - _text) % 512 == 0, "EFI zImage file size is not a multiple of 512 bytes") > >> > >> This would only catch them when EFI is enabled, which doesn't give very > >> good build coverage. I still prefer my solution over adding the assert > >> and a section for _edata - my solution catches it with EFI disabled as > >> well. > >> > >> A variant on that would be this, which should catch additional sections > >> for EFI and non-EFI: > >> > >> diff --git a/arch/arm/boot/compressed/vmlinux.lds.S b/arch/arm/boot/compressed/vmlinux.lds.S > >> index b38dcef90756..9d0c5d80979a 100644 > >> --- a/arch/arm/boot/compressed/vmlinux.lds.S > >> +++ b/arch/arm/boot/compressed/vmlinux.lds.S > >> @@ -95,6 +95,10 @@ SECTIONS > >> > >> _edata = .; > >> > >> + .image_end (NOLOAD) : { > >> + _image_end = .; > >> + } > >> + > >> _magic_sig = ZIMAGE_MAGIC(0x016f2818); > >> _magic_start = ZIMAGE_MAGIC(_start); > >> _magic_end = ZIMAGE_MAGIC(_edata); > >> @@ -119,3 +123,5 @@ SECTIONS > >> .stab.indexstr 0 : { *(.stab.indexstr) } > >> .comment 0 : { *(.comment) } > >> } > >> + > >> +ASSERT(image_end == end, "zImage file size is incorrect") > >> > > > > I take it you mean > > > > _image_end == _edata > > > > here? That works for me: > > > > Tested-by: Ard Biesheuvel > > > > (with CONFIG_EFI enabled) > > > BTW you could simplify that to > > diff --git a/arch/arm/boot/compressed/vmlinux.lds.S > b/arch/arm/boot/compressed/vmlinux.lds.S > index 7a4c59154361..204d4580a04f 100644 > --- a/arch/arm/boot/compressed/vmlinux.lds.S > +++ b/arch/arm/boot/compressed/vmlinux.lds.S > @@ -85,6 +85,10 @@ SECTIONS > > _edata = .; > > + .image_end (NOLOAD) : { > + ASSERT(. == _edata, "zImage file size is incorrect"); > + } Yep. Okay, so now we have a patch from Arnd to fix the alignment problem (so don't need alignment of the piggydata), and we have a patch to catch the additional sections. That leaves a patch to sort out the additional sections. Do you have any preference - I'd prefer one that I can merge along with these changes? One way forward would be to temporarily add the /DISCARD/ for the ksym sections, which can then be removed once your sort() removal gets added to the EFI trees. -- RMK's Patch system: http://www.armlinux.org.uk/developer/patches/ FTTC broadband for 0.8mile line in suburbia: sync at 8.8Mbps down 630kbps up According to speedtest.net: 8.21Mbps down 510kbps up