lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [day] [month] [year] [list]
Message-Id: <20190820072319.2884-1-mpe@ellerman.id.au>
Date:   Tue, 20 Aug 2019 17:23:19 +1000
From:   Michael Ellerman <mpe@...erman.id.au>
To:     will@...nel.org
Cc:     linuxppc-dev@...abs.org, pcc@...gle.com,
        yamada.masahiro@...ionext.com, linux-kernel@...r.kernel.org,
        linux-next@...r.kernel.org
Subject: [PATCH] powerpc/Makefile: Always pass --synthetic to nm if supported

Back in 2004 we added logic to arch/ppc64/Makefile to pass
the --synthetic option to nm, if it was supported by nm.

Then in 2005 when arch/ppc64 and arch/ppc were merged, the logic to
add --synthetic was moved inside an #ifdef CONFIG_PPC64 block within
arch/powerpc/Makefile, and has remained there since.

That was fine, though crufty, until recently when a change to
init/Kconfig added a config time check that uses $(NM). On powerpc
that leads to an infinite loop because Kconfig uses $(NM) to calculate
some values, then the powerpc Makefile changes $(NM), which Kconfig
notices and restarts.

The original commit that added --synthetic simply said:
  On new toolchains we need to use nm --synthetic or we miss code
  symbols.

And the nm man page says that the --synthetic option causes nm to:
  Include synthetic symbols in the output. These are special symbols
  created by the linker for various purposes.

So it seems safe to always pass --synthetic if nm supports it, ie. on
32-bit and 64-bit, it just means 32-bit kernels might have more
symbols reported (and in practice I see no extra symbols). Making it
unconditional avoids the #ifdef CONFIG_PPC64, which in turn avoids the
infinite loop.

Debugged-by: Peter Collingbourne <pcc@...gle.com>
Signed-off-by: Michael Ellerman <mpe@...erman.id.au>
---
 arch/powerpc/Makefile | 2 --
 1 file changed, 2 deletions(-)

See the original discussion here: https://lore.kernel.org/lkml/CAMn1gO6P_VfDRjGZb67ZS4Kh0wjTEQi0cbOkmibTokHQOgP7qw@mail.gmail.com/

 
diff --git a/arch/powerpc/Makefile b/arch/powerpc/Makefile
index c345b79414a9..403f7e193833 100644
--- a/arch/powerpc/Makefile
+++ b/arch/powerpc/Makefile
@@ -39,13 +39,11 @@ endif
 uname := $(shell uname -m)
 KBUILD_DEFCONFIG := $(if $(filter ppc%,$(uname)),$(uname),ppc64)_defconfig
 
-ifdef CONFIG_PPC64
 new_nm := $(shell if $(NM) --help 2>&1 | grep -- '--synthetic' > /dev/null; then echo y; else echo n; fi)
 
 ifeq ($(new_nm),y)
 NM		:= $(NM) --synthetic
 endif
-endif
 
 # BITS is used as extension for files which are available in a 32 bit
 # and a 64 bit version to simplify shared Makefiles.
-- 
2.21.0

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ