+++ Miroslav Benes [11/07/16 16:03 +0200]:

On Mon, 27 Jun 2016, Torsten Duwe wrote:

quoted

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 <http://www.gnu.org/licenses/>.
+ */
+#ifndef _ASM_ARM64_LIVEPATCH_H
+#define _ASM_ARM64_LIVEPATCH_H
+
+#include <linux/module.h>
+#include <linux/ftrace.h>
+
+#ifdef CONFIG_LIVEPATCH

A nit but we removed such guards in the other header files.

quoted

+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.

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(). 

Jessica