From mboxrd@z Thu Jan 1 00:00:00 1970 From: mbenes@suse.cz (Miroslav Benes) Date: Tue, 12 Jul 2016 11:47:41 +0200 (CEST) Subject: arm64: implement live patching In-Reply-To: <20160711215807.GA23097@packer-debian-8-amd64.digitalocean.com> References: <20160627151500.2247A68D20@newverein.lst.de> <20160627151722.B81C968D29@newverein.lst.de> <20160711215807.GA23097@packer-debian-8-amd64.digitalocean.com> Message-ID: To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Mon, 11 Jul 2016, Jessica Yu wrote: > +++ Miroslav Benes [11/07/16 16:03 +0200]: > > On Mon, 27 Jun 2016, Torsten Duwe wrote: > > > > > diff --git a/arch/arm64/include/asm/livepatch.h > > > b/arch/arm64/include/asm/livepatch.h > > > new file mode 100644 > > > index 0000000..6b9a3d1 > > > --- /dev/null > > > +++ b/arch/arm64/include/asm/livepatch.h > > > @@ -0,0 +1,37 @@ > > > +/* > > > + * livepatch.h - arm64-specific Kernel Live Patching Core > > > + * > > > + * Copyright (C) 2016 SUSE > > > + * > > > + * This program is free software; you can redistribute it and/or > > > + * modify it under the terms of the GNU General Public License > > > + * as published by the Free Software Foundation; either version 2 > > > + * of the License, or (at your option) any later version. > > > + * > > > + * This program is distributed in the hope that it will be useful, > > > + * but WITHOUT ANY WARRANTY; without even the implied warranty of > > > + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the > > > + * GNU General Public License for more details. > > > + * > > > + * You should have received a copy of the GNU General Public License > > > + * along with this program; if not, see . > > > + */ > > > +#ifndef _ASM_ARM64_LIVEPATCH_H > > > +#define _ASM_ARM64_LIVEPATCH_H > > > + > > > +#include > > > +#include > > > + > > > +#ifdef CONFIG_LIVEPATCH > > > > A nit but we removed such guards in the other header files. > > > > > +static inline int klp_check_compiler_support(void) > > > +{ > > > + return 0; > > > +} > > > + > > > +static inline void klp_arch_set_pc(struct pt_regs *regs, unsigned long > > > ip) > > > +{ > > > + regs->pc = ip; > > > +} > > > +#endif /* CONFIG_LIVEPATCH */ > > > > I also checked mod_arch_specific structure because of the way we deal > > with relocations. It is defined only if CONFIG_ARM64_MODULE_PLTS is > > enabled and there is a pointer to 'struct elf64_shdr' called plt. It is > > used indirectly in apply_relocate_add() so we need it to stay. However it > > points to an existing Elf section and SHF_ALLOC is added to its sh_flags > > in module_frob_arch_sections() (arch/arm64/kernel/module-plts.c). > > Therefore we should be ok. > > > > Jessica, could you check it as well, please? > > That sounds right, the plt will remain in module core memory, so we > are fine there. > > However I think the plt->sh_size calculation will be incorrect for > livepatch modules. In calculating mod->arch.plt_max_entries (see: > module-plts.c), count_plts() is called for every rela section. > For livepatch modules, this means count_plts() will also be called for > our .klp.rela sections, which is correct behavior. However, > count_plts() only considers relas referring to SHN_UNDEF symbols, and > since every rela in a klp rela section refers to a SHN_LIVEPATCH > symbol, these are all ignored. So count_plts() may return an incorrect > value for a klp rela section. You're right. During the patch module creation we basically transform all SHN_UNDEF relas to SHN_LIVEPATCH, right? We must take it into account here. > Miroslav, can you confirm the issue? I think the fix would be easy > though; we can just add an additional check for SHN_LIVEPATCH in > count_plts(). Yes, such a check should be sufficient. Thanks for looking into it. Miroslav