[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20180313205945.245105-24-thgarnie@google.com>
Date: Tue, 13 Mar 2018 13:59:41 -0700
From: Thomas Garnier <thgarnie@...gle.com>
To: Herbert Xu <herbert@...dor.apana.org.au>,
"David S . Miller" <davem@...emloft.net>,
Thomas Gleixner <tglx@...utronix.de>,
Ingo Molnar <mingo@...hat.com>,
"H . Peter Anvin" <hpa@...or.com>,
Peter Zijlstra <peterz@...radead.org>,
Josh Poimboeuf <jpoimboe@...hat.com>,
Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
Kate Stewart <kstewart@...uxfoundation.org>,
Thomas Garnier <thgarnie@...gle.com>,
Arnd Bergmann <arnd@...db.de>,
Philippe Ombredanne <pombredanne@...b.com>,
Arnaldo Carvalho de Melo <acme@...hat.com>,
Andrey Ryabinin <aryabinin@...tuozzo.com>,
Matthias Kaehlcke <mka@...omium.org>,
Kees Cook <keescook@...omium.org>,
Tom Lendacky <thomas.lendacky@....com>,
"Kirill A . Shutemov" <kirill.shutemov@...ux.intel.com>,
Andy Lutomirski <luto@...nel.org>,
Dominik Brodowski <linux@...inikbrodowski.net>,
Borislav Petkov <bp@...en8.de>,
Borislav Petkov <bp@...e.de>,
"Rafael J . Wysocki" <rjw@...ysocki.net>,
Len Brown <len.brown@...el.com>,
Pavel Machek <pavel@....cz>,
Juergen Gross <jgross@...e.com>,
Alok Kataria <akataria@...are.com>,
Steven Rostedt <rostedt@...dmis.org>,
Tejun Heo <tj@...nel.org>,
Christoph Lameter <cl@...ux.com>,
Dennis Zhou <dennisszhou@...il.com>,
Boris Ostrovsky <boris.ostrovsky@...cle.com>,
David Woodhouse <dwmw@...zon.co.uk>,
Alexey Dobriyan <adobriyan@...il.com>,
"Paul E . McKenney" <paulmck@...ux.vnet.ibm.com>,
Andrew Morton <akpm@...ux-foundation.org>,
Nicolas Pitre <nicolas.pitre@...aro.org>,
Randy Dunlap <rdunlap@...radead.org>,
"Luis R . Rodriguez" <mcgrof@...nel.org>,
Christopher Li <sparse@...isli.org>,
Jason Baron <jbaron@...mai.com>,
Ashish Kalra <ashish@...estacks.com>,
Kyle McMartin <kyle@...hat.com>,
Dou Liyang <douly.fnst@...fujitsu.com>,
Lukas Wunner <lukas@...ner.de>,
Petr Mladek <pmladek@...e.com>,
Sergey Senozhatsky <sergey.senozhatsky.work@...il.com>,
Masahiro Yamada <yamada.masahiro@...ionext.com>,
Ingo Molnar <mingo@...nel.org>,
Nicholas Piggin <npiggin@...il.com>,
Cao jin <caoj.fnst@...fujitsu.com>,
"H . J . Lu" <hjl.tools@...il.com>,
Paolo Bonzini <pbonzini@...hat.com>,
Radim Krčmář <rkrcmar@...hat.com>,
Joerg Roedel <joro@...tes.org>,
Dave Hansen <dave.hansen@...ux.intel.com>,
Rik van Riel <riel@...hat.com>,
Jia Zhang <qianyue.zj@...baba-inc.com>,
Jiri Slaby <jslaby@...e.cz>,
Kyle Huey <me@...ehuey.com>,
Jonathan Corbet <corbet@....net>,
Matthew Wilcox <mawilcox@...rosoft.com>,
Michal Hocko <mhocko@...e.com>,
Rob Landley <rob@...dley.net>,
Baoquan He <bhe@...hat.com>,
Daniel Micay <danielmicay@...il.com>,
Jan H . Schönherr <jschoenh@...zon.de>
Cc: x86@...nel.org,
linux-crypto@...r.kernel.org,
linux-kernel@...r.kernel.org,
linux-pm@...r.kernel.org,
virtualization@...ts.linux-foundation.org,
xen-devel@...ts.xenproject.org,
linux-arch@...r.kernel.org,
linux-sparse@...r.kernel.org,
kvm@...r.kernel.org,
linux-doc@...r.kernel.org,
kernel-hardening@...ts.openwall.com
Subject: [PATCH v2 23/27] x86/modules: Adapt module loading for PIE support
Adapt module loading to support PIE relocations. Generate dynamic GOT if
a symbol requires it but no entry exist in the kernel GOT.
Position Independent Executable (PIE) support will allow to extended the
KASLR randomization range below the -2G memory limit.
Signed-off-by: Thomas Garnier <thgarnie@...gle.com>
---
arch/x86/Makefile | 4 +
arch/x86/include/asm/module.h | 11 ++
arch/x86/include/asm/sections.h | 4 +
arch/x86/kernel/module.c | 181 +++++++++++++++++++++++++++++++-
arch/x86/kernel/module.lds | 3 +
5 files changed, 198 insertions(+), 5 deletions(-)
create mode 100644 arch/x86/kernel/module.lds
diff --git a/arch/x86/Makefile b/arch/x86/Makefile
index 16dafc551f3b..f24d200c0d9d 100644
--- a/arch/x86/Makefile
+++ b/arch/x86/Makefile
@@ -135,7 +135,11 @@ else
KBUILD_CFLAGS += $(cflags-y)
KBUILD_CFLAGS += -mno-red-zone
+ifdef CONFIG_X86_PIE
+ KBUILD_LDFLAGS_MODULE += -T $(srctree)/arch/x86/kernel/module.lds
+else
KBUILD_CFLAGS += -mcmodel=kernel
+endif
# -funit-at-a-time shrinks the kernel .text considerably
# unfortunately it makes reading oopses harder.
diff --git a/arch/x86/include/asm/module.h b/arch/x86/include/asm/module.h
index 7948a17febb4..68ff05e14288 100644
--- a/arch/x86/include/asm/module.h
+++ b/arch/x86/include/asm/module.h
@@ -5,12 +5,23 @@
#include <asm-generic/module.h>
#include <asm/orc_types.h>
+#ifdef CONFIG_X86_PIE
+struct mod_got_sec {
+ struct elf64_shdr *got;
+ int got_num_entries;
+ int got_max_entries;
+};
+#endif
+
struct mod_arch_specific {
#ifdef CONFIG_UNWINDER_ORC
unsigned int num_orcs;
int *orc_unwind_ip;
struct orc_entry *orc_unwind;
#endif
+#ifdef CONFIG_X86_PIE
+ struct mod_got_sec core;
+#endif
};
#ifdef CONFIG_X86_64
diff --git a/arch/x86/include/asm/sections.h b/arch/x86/include/asm/sections.h
index cad292f62eed..0bbd9f941573 100644
--- a/arch/x86/include/asm/sections.h
+++ b/arch/x86/include/asm/sections.h
@@ -16,4 +16,8 @@ extern char __end_rodata_hpage_align[];
extern char __start_got[], __end_got[];
#endif
+#if defined(CONFIG_X86_PIE)
+extern char __start_got[], __end_got[];
+#endif
+
#endif /* _ASM_X86_SECTIONS_H */
diff --git a/arch/x86/kernel/module.c b/arch/x86/kernel/module.c
index f58336af095c..88895f3d474b 100644
--- a/arch/x86/kernel/module.c
+++ b/arch/x86/kernel/module.c
@@ -30,6 +30,7 @@
#include <linux/gfp.h>
#include <linux/jump_label.h>
#include <linux/random.h>
+#include <linux/sort.h>
#include <asm/text-patching.h>
#include <asm/page.h>
@@ -77,6 +78,173 @@ static unsigned long int get_module_load_offset(void)
}
#endif
+#ifdef CONFIG_X86_PIE
+static u64 find_got_kernel_entry(Elf64_Sym *sym, const Elf64_Rela *rela)
+{
+ u64 *pos;
+
+ for (pos = (u64*)__start_got; pos < (u64*)__end_got; pos++) {
+ if (*pos == sym->st_value)
+ return (u64)pos + rela->r_addend;
+ }
+
+ return 0;
+}
+
+static u64 module_emit_got_entry(struct module *mod, void *loc,
+ const Elf64_Rela *rela, Elf64_Sym *sym)
+{
+ struct mod_got_sec *gotsec = &mod->arch.core;
+ u64 *got = (u64*)gotsec->got->sh_addr;
+ int i = gotsec->got_num_entries;
+ u64 ret;
+
+ /* Check if we can use the kernel GOT */
+ ret = find_got_kernel_entry(sym, rela);
+ if (ret)
+ return ret;
+
+ got[i] = sym->st_value;
+
+ /*
+ * Check if the entry we just created is a duplicate. Given that the
+ * relocations are sorted, this will be the last entry we allocated.
+ * (if one exists).
+ */
+ if (i > 0 && got[i] == got[i - 2]) {
+ ret = (u64)&got[i - 1];
+ } else {
+ gotsec->got_num_entries++;
+ BUG_ON(gotsec->got_num_entries > gotsec->got_max_entries);
+ ret = (u64)&got[i];
+ }
+
+ return ret + rela->r_addend;
+}
+
+#define cmp_3way(a,b) ((a) < (b) ? -1 : (a) > (b))
+
+static int cmp_rela(const void *a, const void *b)
+{
+ const Elf64_Rela *x = a, *y = b;
+ int i;
+
+ /* sort by type, symbol index and addend */
+ i = cmp_3way(ELF64_R_TYPE(x->r_info), ELF64_R_TYPE(y->r_info));
+ if (i == 0)
+ i = cmp_3way(ELF64_R_SYM(x->r_info), ELF64_R_SYM(y->r_info));
+ if (i == 0)
+ i = cmp_3way(x->r_addend, y->r_addend);
+ return i;
+}
+
+static bool duplicate_rel(const Elf64_Rela *rela, int num)
+{
+ /*
+ * Entries are sorted by type, symbol index and addend. That means
+ * that, if a duplicate entry exists, it must be in the preceding
+ * slot.
+ */
+ return num > 0 && cmp_rela(rela + num, rela + num - 1) == 0;
+}
+
+static unsigned int count_gots(Elf64_Sym *syms, Elf64_Rela *rela, int num)
+{
+ unsigned int ret = 0;
+ Elf64_Sym *s;
+ int i;
+
+ for (i = 0; i < num; i++) {
+ switch (ELF64_R_TYPE(rela[i].r_info)) {
+ case R_X86_64_GOTPCREL:
+ s = syms + ELF64_R_SYM(rela[i].r_info);
+
+ /*
+ * Use the kernel GOT when possible, else reserve a
+ * custom one for this module.
+ */
+ if (!duplicate_rel(rela, i) &&
+ !find_got_kernel_entry(s, rela + i))
+ ret++;
+ break;
+ }
+ }
+ return ret;
+}
+
+/*
+ * Generate GOT entries for GOTPCREL relocations that do not exists in the
+ * kernel GOT. Based on arm64 module-plts implementation.
+ */
+int module_frob_arch_sections(Elf_Ehdr *ehdr, Elf_Shdr *sechdrs,
+ char *secstrings, struct module *mod)
+{
+ unsigned long gots = 0;
+ Elf_Shdr *symtab = NULL;
+ Elf64_Sym *syms = NULL;
+ char *strings, *name;
+ int i;
+
+ /*
+ * Find the empty .got section so we can expand it to store the PLT
+ * entries. Record the symtab address as well.
+ */
+ for (i = 0; i < ehdr->e_shnum; i++) {
+ if (!strcmp(secstrings + sechdrs[i].sh_name, ".got")) {
+ mod->arch.core.got = sechdrs + i;
+ } else if (sechdrs[i].sh_type == SHT_SYMTAB) {
+ symtab = sechdrs + i;
+ syms = (Elf64_Sym *)symtab->sh_addr;
+ }
+ }
+
+ if (!mod->arch.core.got) {
+ pr_err("%s: module GOT section missing\n", mod->name);
+ return -ENOEXEC;
+ }
+ if (!syms) {
+ pr_err("%s: module symtab section missing\n", mod->name);
+ return -ENOEXEC;
+ }
+
+ for (i = 0; i < ehdr->e_shnum; i++) {
+ Elf64_Rela *rels = (void *)ehdr + sechdrs[i].sh_offset;
+ int numrels = sechdrs[i].sh_size / sizeof(Elf64_Rela);
+
+ if (sechdrs[i].sh_type != SHT_RELA)
+ continue;
+
+ /* sort by type, symbol index and addend */
+ sort(rels, numrels, sizeof(Elf64_Rela), cmp_rela, NULL);
+
+ gots += count_gots(syms, rels, numrels);
+ }
+
+ mod->arch.core.got->sh_type = SHT_NOBITS;
+ mod->arch.core.got->sh_flags = SHF_ALLOC;
+ mod->arch.core.got->sh_addralign = L1_CACHE_BYTES;
+ mod->arch.core.got->sh_size = (gots + 1) * sizeof(u64);
+ mod->arch.core.got_num_entries = 0;
+ mod->arch.core.got_max_entries = gots;
+
+ /*
+ * If a _GLOBAL_OFFSET_TABLE_ symbol exists, make it absolute for
+ * modules to correctly reference it. Similar to s390 implementation.
+ */
+ strings = (void *) ehdr + sechdrs[symtab->sh_link].sh_offset;
+ for (i = 0; i < symtab->sh_size/sizeof(Elf_Sym); i++) {
+ if (syms[i].st_shndx != SHN_UNDEF)
+ continue;
+ name = strings + syms[i].st_name;
+ if (!strcmp(name, "_GLOBAL_OFFSET_TABLE_")) {
+ syms[i].st_shndx = SHN_ABS;
+ break;
+ }
+ }
+ return 0;
+}
+#endif
+
void *module_alloc(unsigned long size)
{
void *p;
@@ -190,16 +358,20 @@ int apply_relocate_add(Elf64_Shdr *sechdrs,
if ((s64)val != *(s32 *)loc)
goto overflow;
break;
+#ifdef CONFIG_X86_PIE
+ case R_X86_64_GOTPCREL:
+ val = module_emit_got_entry(me, loc, rel + i, sym);
+ /* fallthrough */
+#endif
case R_X86_64_PC32:
case R_X86_64_PLT32:
if (*(u32 *)loc != 0)
goto invalid_relocation;
val -= (u64)loc;
*(u32 *)loc = val;
-#if 0
- if ((s64)val != *(s32 *)loc)
+ if (IS_ENABLED(CONFIG_X86_PIE) &&
+ (s64)val != *(s32 *)loc)
goto overflow;
-#endif
break;
default:
pr_err("%s: Unknown rela relocation: %llu\n",
@@ -217,8 +389,7 @@ int apply_relocate_add(Elf64_Shdr *sechdrs,
overflow:
pr_err("overflow in relocation type %d val %Lx\n",
(int)ELF64_R_TYPE(rel[i].r_info), val);
- pr_err("`%s' likely not compiled with -mcmodel=kernel\n",
- me->name);
+ pr_err("`%s' likely too far from the kernel\n", me->name);
return -ENOEXEC;
}
#endif
diff --git a/arch/x86/kernel/module.lds b/arch/x86/kernel/module.lds
new file mode 100644
index 000000000000..fd6e95a4b454
--- /dev/null
+++ b/arch/x86/kernel/module.lds
@@ -0,0 +1,3 @@
+SECTIONS {
+ .got (NOLOAD) : { BYTE(0) }
+}
--
2.16.2.660.g709887971b-goog
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
