All of lore.kernel.org
 help / color / mirror / Atom feed
From: Nick Desaulniers <ndesaulniers@google.com>
To: Masahiro Yamada <yamada.masahiro@socionext.com>
Cc: Linux Kbuild mailing list <linux-kbuild@vger.kernel.org>,
	Michael Davidson <md@google.com>,
	Greg Hackmann <ghackmann@google.com>,
	Pirama Arumuga Nainar <pirama@google.com>,
	Douglas Anderson <dianders@chromium.org>,
	Ingo Molnar <mingo@kernel.org>,
	Matthias Kaehlcke <mka@chromium.org>,
	Arnd Bergmann <arnd@arndb.de>,
	Marcin Nowakowski <marcin.nowakowski@imgtec.com>,
	Mark Charlebois <charlebm@gmail.com>,
	Josh Poimboeuf <jpoimboe@redhat.com>,
	Cao jin <caoj.fnst@cn.fujitsu.com>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH v2] kbuild: fix linker feature test macros when cross compiling with Clang
Date: Mon, 6 Nov 2017 10:47:37 -0800	[thread overview]
Message-ID: <CAKwvOdmhuaLC6=vG2DMFAbhfXoUr74_aemM2DK7me6pQ0obW4g@mail.gmail.com> (raw)
In-Reply-To: <CAKwvOdn1q=9E+dHH_mX+GNsoVjozfGMjGVULG32935vJgjDyuA@mail.gmail.com>

Comparing make V=1 with the suggested config before my patch, after my
patch, and after Masahiro's suggestion to add $(LDFLAGS):

before:
...
ld -m elf_i386  -pie    -T arch/x86/boot/compressed/vmlinux.lds ...
...

after my:
...
ld -m elf_i386     -T arch/x86/boot/compressed/vmlinux.lds ...
...

after my+Masahiro:
...
ld -m elf_i386 -pie    -T arch/x86/boot/compressed/vmlinux.lds ...
...

:)

Just to dig into this a little more (though I suspect we've found the issue):

Next is to debug what we're passing to try-run and see what errors
it's masking.  Adding to arch/x86/boot/compressed/Makefile:
--- a/arch/x86/boot/compressed/Makefile
+++ b/arch/x86/boot/compressed/Makefile
@@ -44,6 +44,10 @@ LDFLAGS := -m elf_$(UTS_MACHINE)
 # Compressed kernel should be built as PIE since it may be loaded at any
 # address by the bootloader.
 ifeq ($(CONFIG_X86_32),y)
+$(info "XXX")
+$(info "KBUILD_CPPFLAGS: ${KBUILD_CPPFLAGS}")
+$(info "CC_OPTION_CFLAGS: ${CC_OPTION_CFLAGS}")
+$(info "LDFLAGS: ${LDFLAGS}")
 LDFLAGS += $(call ld-option, -pie) $(call ld-option, --no-dynamic-linker)
 else
 # To build 64-bit compressed kernel as PIE, we disable relocation

then grepping for XXX in a build output:
"XXX"
"KBUILD_CPPFLAGS: -D__KERNEL__  "
"CC_OPTION_CFLAGS: -m32 -D__KERNEL__ -O2 -fno-strict-aliasing -fPIE
-DDISABLE_BRANCH_PROFILING -march=i386 -mno-mmx -mno-sse
-ffreestanding -fno-stack-protector"
"LD_FLAGS: "

then trying this on the command line:
➜  kernel-all git:(kbuild) ✗ cc -pie -D__KERNEL__ -m32 -D__KERNEL__
-O2 -fno-strict-aliasing -fPIE -DDISABLE_BRANCH_PROFILING -march=i386
-mno-mmx -mno-sse -ffreestanding -fno-stack-protector -x c /dev/null
-c -o temp.o
➜  kernel-all git:(kbuild) ✗ ld -pie temp.o -o temp
ld: i386 architecture of input file `temp.o' is incompatible with
i386:x86-64 output
ld: warning: cannot find entry symbol _start; defaulting to 0000000000000078
➜  kernel-all git:(kbuild) ✗ echo $?
1
➜  kernel-all git:(kbuild) ✗ file temp.o
temp.o: ELF 32-bit LSB  relocatable, Intel 80386, version 1 (SYSV), not stripped

So it looks like without LDFLAGS, we don't tell the linker to link as
32b, causing ld-option to fail to add -pie to LDFLAGS.

➜  kernel-all git:(kbuild) ✗ ld -pie -m elf_i386 temp.o -o temp
ld: warning: cannot find entry symbol _start; defaulting to 0000000000000185
➜  kernel-all git:(kbuild) ✗ echo $?
0

sure enough, just before the call to ld-option from
arch/x86/boot/compressed/Makefile:

 43 LDFLAGS := -m elf_$(UTS_MACHINE)

so looks like Masahiro's suggestion is correct.  Will send a v3.

  reply	other threads:[~2017-11-06 18:47 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-10-26 20:17 [PATCH] kbuild: fix linker feature test macros when cross compiling with Clang Nick Desaulniers
2017-10-27 11:20 ` Masahiro Yamada
2017-10-27 18:28   ` Nick Desaulniers
2017-10-27 20:10     ` Nick Desaulniers
2017-10-28 14:59     ` Masahiro Yamada
2017-10-27 20:13   ` [PATCH v2] " Nick Desaulniers
2017-10-28 15:00     ` Masahiro Yamada
2017-10-30  6:50       ` Masahiro Yamada
2017-10-30 15:46         ` Masahiro Yamada
2017-10-30 16:13           ` Nick Desaulniers
2017-11-06 18:47             ` Nick Desaulniers [this message]
2017-11-06 18:47               ` [PATCH v3] " Nick Desaulniers
2017-11-07  3:41                 ` Masahiro Yamada
2017-12-11 10:17                 ` Arnd Bergmann
2017-12-11 11:47                   ` Masahiro Yamada

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CAKwvOdmhuaLC6=vG2DMFAbhfXoUr74_aemM2DK7me6pQ0obW4g@mail.gmail.com' \
    --to=ndesaulniers@google.com \
    --cc=arnd@arndb.de \
    --cc=caoj.fnst@cn.fujitsu.com \
    --cc=charlebm@gmail.com \
    --cc=dianders@chromium.org \
    --cc=ghackmann@google.com \
    --cc=jpoimboe@redhat.com \
    --cc=linux-kbuild@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=marcin.nowakowski@imgtec.com \
    --cc=md@google.com \
    --cc=mingo@kernel.org \
    --cc=mka@chromium.org \
    --cc=pirama@google.com \
    --cc=yamada.masahiro@socionext.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.