From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx0a-001b2d01.pphosted.com (mx0b-001b2d01.pphosted.com [148.163.158.5]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by lists.ozlabs.org (Postfix) with ESMTPS id 3ysSw54TlQzDrvV for ; Thu, 7 Dec 2017 06:14:13 +1100 (AEDT) Received: from pps.filterd (m0098416.ppops.net [127.0.0.1]) by mx0b-001b2d01.pphosted.com (8.16.0.21/8.16.0.21) with SMTP id vB6JE9TE024372 for ; Wed, 6 Dec 2017 14:14:09 -0500 Received: from e35.co.us.ibm.com (e35.co.us.ibm.com [32.97.110.153]) by mx0b-001b2d01.pphosted.com with ESMTP id 2epm4y6nya-1 (version=TLSv1.2 cipher=AES256-SHA bits=256 verify=NOT) for ; Wed, 06 Dec 2017 14:14:09 -0500 Received: from localhost by e35.co.us.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Wed, 6 Dec 2017 12:14:05 -0700 From: Desnes Augusto Nunes do Rosario To: linuxppc-dev@lists.ozlabs.org, linux-kernel@vger.kernel.org, amodra@gmail.com Cc: desnesn@linux.vnet.ibm.com, benh@kernel.crashing.org, paulus@samba.org, mpe@ellerman.id.au, ard.biesheuvel@linaro.org, rusty@rustcorp.com.au Subject: [PATCH] [powerpc-next] Fix powerpc64 alignment of .toc section in kernel modules Date: Wed, 6 Dec 2017 17:12:28 -0200 Message-Id: <20171206191228.30830-1-desnesn@linux.vnet.ibm.com> List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , powerpc64 gcc can generate code that offsets an address, to access part of an object in memory. If the address is a -mcmodel=medium toc pointer relative address then code like the following is possible. addis r9,r2,var@toc@ha ld r3,var@toc@l(r9) ld r4,(var+8)@toc@l(r9) This works fine so long as var is naturally aligned, *and* r2 is sufficiently aligned. If not, there is a possibility that the offset added to access var+8 wraps over a n*64k+32k boundary. Modules don't have any guarantee that r2 is sufficiently aligned. Moreover, code generated by older compilers generates a .toc section with 2**0 alignment, which can result in relocation failures at module load time even without the wrap problem. Thus, this patch links modules with an aligned .toc section (Makefile and module.lds changes), and forces alignment for out of tree modules or those without a .toc section (module_64.c changes). Signed-off-by: Alan Modra [ desnesn: updated patch to apply to powerpc-next kernel v4.15 ] Signed-off-by: Desnes A. Nunes do Rosario --- arch/powerpc/Makefile | 1 + arch/powerpc/kernel/module.lds | 8 ++++++++ arch/powerpc/kernel/module_64.c | 13 +++++++++---- 3 files changed, 18 insertions(+), 4 deletions(-) create mode 100644 arch/powerpc/kernel/module.lds diff --git a/arch/powerpc/Makefile b/arch/powerpc/Makefile index 1381693..c472f5b 100644 --- a/arch/powerpc/Makefile +++ b/arch/powerpc/Makefile @@ -63,6 +63,7 @@ UTS_MACHINE := $(subst $(space),,$(machine-y)) ifdef CONFIG_PPC32 KBUILD_LDFLAGS_MODULE += arch/powerpc/lib/crtsavres.o else +KBUILD_LDFLAGS_MODULE += -T arch/powerpc/kernel/module.lds ifeq ($(call ld-ifversion, -ge, 225000000, y),y) # Have the linker provide sfpr if possible. # There is a corresponding test in arch/powerpc/lib/Makefile diff --git a/arch/powerpc/kernel/module.lds b/arch/powerpc/kernel/module.lds new file mode 100644 index 0000000..cea5dc1 --- /dev/null +++ b/arch/powerpc/kernel/module.lds @@ -0,0 +1,8 @@ +/* Force alignment of .toc section. */ +SECTIONS +{ + .toc 0 : ALIGN(256) + { + *(.got .toc) + } +} diff --git a/arch/powerpc/kernel/module_64.c b/arch/powerpc/kernel/module_64.c index 759104b..9b2c5c1 100644 --- a/arch/powerpc/kernel/module_64.c +++ b/arch/powerpc/kernel/module_64.c @@ -339,8 +339,11 @@ int module_frob_arch_sections(Elf64_Ehdr *hdr, char *p; if (strcmp(secstrings + sechdrs[i].sh_name, ".stubs") == 0) me->arch.stubs_section = i; - else if (strcmp(secstrings + sechdrs[i].sh_name, ".toc") == 0) + else if (strcmp(secstrings + sechdrs[i].sh_name, ".toc") == 0) { me->arch.toc_section = i; + if (sechdrs[i].sh_addralign < 8) + sechdrs[i].sh_addralign = 8; + } else if (strcmp(secstrings+sechdrs[i].sh_name,"__versions")==0) dedotify_versions((void *)hdr + sechdrs[i].sh_offset, sechdrs[i].sh_size); @@ -374,11 +377,13 @@ int module_frob_arch_sections(Elf64_Ehdr *hdr, } /* r2 is the TOC pointer: it actually points 0x8000 into the TOC (this - gives the value maximum span in an instruction which uses a signed - offset) */ + * gives the value maximum span in an instruction which uses a signed + * offset). Round down to a 256 byte boundary for the odd case where + * we are setting up r2 without a .toc section. + */ static inline unsigned long my_r2(const Elf64_Shdr *sechdrs, struct module *me) { - return sechdrs[me->arch.toc_section].sh_addr + 0x8000; + return (sechdrs[me->arch.toc_section].sh_addr & -256) + 0x8000; } /* Both low and high 16 bits are added as SIGNED additions, so if low -- 2.9.5