Skip to contentSkip to conversation

arm64: Fix text patching logic when using fixmap

PATCH New Marc Zyngier 6 participants 7 messages Feb 24, 2015 – Feb 27, 2015 arm64

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 arm64: Fix text patching logic when using fixmap

Patch 2f896d586610 ("arm64: use fixmap for text patching") changed the way we patch the kernel text, using a fixmap when the kernel or modules are flagged as read only.

Unfortunately, a flaw in the logic makes it fall over when patching modules without CONFIG_DEBUG_SET_MODULE_RONX enabled:

[...]

[   32.032636] Call trace:
[   32.032716] [<fffffe00003da0dc>] __copy_to_user+0x2c/0x60
[   32.032837] [<fffffe0000099f08>] __aarch64_insn_write+0x94/0xf8
[   32.033027] [<fffffe000009a0a0>] aarch64_insn_patch_text_nosync+0x18/0x58
[   32.033200] [<fffffe000009c3ec>] ftrace_modify_code+0x58/0x84
[   32.033363] [<fffffe000009c4e4>] ftrace_make_nop+0x3c/0x58
[   32.033532] [<fffffe0000164420>] ftrace_process_locs+0x3d0/0x5c8
[   32.033709] [<fffffe00001661cc>] ftrace_module_init+0x28/0x34
[   32.033882] [<fffffe0000135148>] load_module+0xbb8/0xfc4
[   32.034044] [<fffffe0000135714>] SyS_finit_module+0x94/0xc4

[...]

This is triggered by the use of virt_to_page() on a module address, which ends to pointing to Nowhereland if you're lucky, or corrupt your precious data if not.

This patch fixes the logic by mimicking what is done on arm:
- If we're patching a module and CONFIG_DEBUG_SET_MODULE_RONX is set, use vmalloc_to_page().
- If we're patching the kernel and CONFIG_DEBUG_RODATA is set, use virt_to_page().
- Otherwise, use the provided address, as we can write to it directly.

Tested on 4.0-rc1 as a KVM guest.

Cc Kees Cook <[email protected]>
Cc Mark Rutland <[email protected]>
Cc Laura Abbott <[email protected]>
Cc Catalin Marinas <[email protected]>
Cc Will Deacon <[email protected]>
---
 arch/arm64/kernel/insn.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
arch/arm64/kernel/insn.c +3 -1
index 27d4864..c8eca88 100644
@@ -87,8 +87,10 @@ static void __kprobes *patch_map(void *addr, int fixmap)
 
 	if (module && IS_ENABLED(CONFIG_DEBUG_SET_MODULE_RONX))
 		page = vmalloc_to_page(addr);
-	else
+	else if (!module && IS_ENABLED(CONFIG_DEBUG_RODATA))
 		page = virt_to_page(addr);
+	else
+		return addr;
 
 	BUG_ON(!page);
 	set_fixmap(fixmap, page_to_phys(page));
Signature · 1 line
2.1.4
#1 lore raw

PATCH 1/1 arm64: Fix text patching logic when using fixmap

Quoted 33 lines · Marc Zyngier

Patch 2f896d586610 ("arm64: use fixmap for text patching") changed the way we patch the kernel text, using a fixmap when the kernel or modules are flagged as read only.

Unfortunately, a flaw in the logic makes it fall over when patching modules without CONFIG_DEBUG_SET_MODULE_RONX enabled:

[...]

[   32.032636] Call trace:
[   32.032716] [<fffffe00003da0dc>] __copy_to_user+0x2c/0x60
[   32.032837] [<fffffe0000099f08>] __aarch64_insn_write+0x94/0xf8
[   32.033027] [<fffffe000009a0a0>] aarch64_insn_patch_text_nosync+0x18/0x58
[   32.033200] [<fffffe000009c3ec>] ftrace_modify_code+0x58/0x84
[   32.033363] [<fffffe000009c4e4>] ftrace_make_nop+0x3c/0x58
[   32.033532] [<fffffe0000164420>] ftrace_process_locs+0x3d0/0x5c8
[   32.033709] [<fffffe00001661cc>] ftrace_module_init+0x28/0x34
[   32.033882] [<fffffe0000135148>] load_module+0xbb8/0xfc4
[   32.034044] [<fffffe0000135714>] SyS_finit_module+0x94/0xc4

[...]

This is triggered by the use of virt_to_page() on a module address, which ends to pointing to Nowhereland if you're lucky, or corrupt your precious data if not.

This patch fixes the logic by mimicking what is done on arm:
- If we're patching a module and CONFIG_DEBUG_SET_MODULE_RONX is set, use vmalloc_to_page().
- If we're patching the kernel and CONFIG_DEBUG_RODATA is set, use virt_to_page().
- Otherwise, use the provided address, as we can write to it directly.

Tested on 4.0-rc1 as a KVM guest.

Quoted 7 lines
Cc Kees Cook <[email protected]>
Cc Mark Rutland <[email protected]>
Cc Laura Abbott <[email protected]>
Cc Catalin Marinas <[email protected]>
Cc Will Deacon <[email protected]>

Good catch! Looks fine to me.

-Kees

Quoted 23 lines
---
 arch/arm64/kernel/insn.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
arch/arm64/kernel/insn.c +3 -2
index 27d4864..c8eca88 100644
@@ -87,8 +87,10 @@ static void __kprobes *patch_map(void *addr, int fixmap)
 
        if (module && IS_ENABLED(CONFIG_DEBUG_SET_MODULE_RONX))
                page = vmalloc_to_page(addr);
-       else
+       else if (!module && IS_ENABLED(CONFIG_DEBUG_RODATA))
                page = virt_to_page(addr);
+       else
+               return addr;
 
        BUG_ON(!page);
        set_fixmap(fixmap, page_to_phys(page));
--

2.1.4

Signature · 2 lines
Kees Cook
Chrome OS Security
reviewer
#2 lore raw

PATCH 1/1 arm64: Fix text patching logic when using fixmap

In reply to Marc Zyngier
Quoted 33 lines · Marc Zyngier

Patch 2f896d586610 ("arm64: use fixmap for text patching") changed the way we patch the kernel text, using a fixmap when the kernel or modules are flagged as read only.

Unfortunately, a flaw in the logic makes it fall over when patching modules without CONFIG_DEBUG_SET_MODULE_RONX enabled:

[...]

[   32.032636] Call trace:
[   32.032716] [<fffffe00003da0dc>] __copy_to_user+0x2c/0x60
[   32.032837] [<fffffe0000099f08>] __aarch64_insn_write+0x94/0xf8
[   32.033027] [<fffffe000009a0a0>] aarch64_insn_patch_text_nosync+0x18/0x58
[   32.033200] [<fffffe000009c3ec>] ftrace_modify_code+0x58/0x84
[   32.033363] [<fffffe000009c4e4>] ftrace_make_nop+0x3c/0x58
[   32.033532] [<fffffe0000164420>] ftrace_process_locs+0x3d0/0x5c8
[   32.033709] [<fffffe00001661cc>] ftrace_module_init+0x28/0x34
[   32.033882] [<fffffe0000135148>] load_module+0xbb8/0xfc4
[   32.034044] [<fffffe0000135714>] SyS_finit_module+0x94/0xc4

[...]

This is triggered by the use of virt_to_page() on a module address, which ends to pointing to Nowhereland if you're lucky, or corrupt your precious data if not.

This patch fixes the logic by mimicking what is done on arm:
- If we're patching a module and CONFIG_DEBUG_SET_MODULE_RONX is set, use vmalloc_to_page().
- If we're patching the kernel and CONFIG_DEBUG_RODATA is set, use virt_to_page().
- Otherwise, use the provided address, as we can write to it directly.

Tested on 4.0-rc1 as a KVM guest.

Quoted 7 lines
Cc Kees Cook <[email protected]>
Cc Mark Rutland <[email protected]>
Cc Laura Abbott <[email protected]>
Cc Catalin Marinas <[email protected]>
Cc Will Deacon <[email protected]>

This makes sense to me:

Mark.

Quoted 24 lines
---
 arch/arm64/kernel/insn.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
arch/arm64/kernel/insn.c +3 -1
index 27d4864..c8eca88 100644
@@ -87,8 +87,10 @@ static void __kprobes *patch_map(void *addr, int fixmap)
 
 	if (module && IS_ENABLED(CONFIG_DEBUG_SET_MODULE_RONX))
 		page = vmalloc_to_page(addr);
-	else
+	else if (!module && IS_ENABLED(CONFIG_DEBUG_RODATA))
 		page = virt_to_page(addr);
+	else
+		return addr;
 
 	BUG_ON(!page);
 	set_fixmap(fixmap, page_to_phys(page));
Signature · 1 line
2.1.4
#3 lore raw

PATCH 1/1 arm64: Fix text patching logic when using fixmap

In reply to Marc Zyngier
Quoted 33 lines · Marc Zyngier

Patch 2f896d586610 ("arm64: use fixmap for text patching") changed the way we patch the kernel text, using a fixmap when the kernel or modules are flagged as read only.

Unfortunately, a flaw in the logic makes it fall over when patching modules without CONFIG_DEBUG_SET_MODULE_RONX enabled:

[...]

[   32.032636] Call trace:
[   32.032716] [<fffffe00003da0dc>] __copy_to_user+0x2c/0x60
[   32.032837] [<fffffe0000099f08>] __aarch64_insn_write+0x94/0xf8
[   32.033027] [<fffffe000009a0a0>] aarch64_insn_patch_text_nosync+0x18/0x58
[   32.033200] [<fffffe000009c3ec>] ftrace_modify_code+0x58/0x84
[   32.033363] [<fffffe000009c4e4>] ftrace_make_nop+0x3c/0x58
[   32.033532] [<fffffe0000164420>] ftrace_process_locs+0x3d0/0x5c8
[   32.033709] [<fffffe00001661cc>] ftrace_module_init+0x28/0x34
[   32.033882] [<fffffe0000135148>] load_module+0xbb8/0xfc4
[   32.034044] [<fffffe0000135714>] SyS_finit_module+0x94/0xc4

[...]

This is triggered by the use of virt_to_page() on a module address, which ends to pointing to Nowhereland if you're lucky, or corrupt your precious data if not.

This patch fixes the logic by mimicking what is done on arm:
- If we're patching a module and CONFIG_DEBUG_SET_MODULE_RONX is set, use vmalloc_to_page().
- If we're patching the kernel and CONFIG_DEBUG_RODATA is set, use virt_to_page().
- Otherwise, use the provided address, as we can write to it directly.

Tested on 4.0-rc1 as a KVM guest.

Quoted 7 lines
Cc Kees Cook <[email protected]>
Cc Mark Rutland <[email protected]>
Cc Laura Abbott <[email protected]>
Cc Catalin Marinas <[email protected]>
Cc Will Deacon <[email protected]>

Tested and works for me, thanks! Therefore:

Rich.

Quoted 21 lines
 arch/arm64/kernel/insn.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
arch/arm64/kernel/insn.c +3 -1
index 27d4864..c8eca88 100644
@@ -87,8 +87,10 @@ static void __kprobes *patch_map(void *addr, int fixmap)
 
 	if (module && IS_ENABLED(CONFIG_DEBUG_SET_MODULE_RONX))
 		page = vmalloc_to_page(addr);
-	else
+	else if (!module && IS_ENABLED(CONFIG_DEBUG_RODATA))
 		page = virt_to_page(addr);
+	else
+		return addr;
 
 	BUG_ON(!page);
 	set_fixmap(fixmap, page_to_phys(page));
Signature · 1 line
2.1.4
Signature · 5 lines
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-p2v converts physical machines to virtual machines.  Boot with a
live CD or over the network (PXE) and turn machines into KVM guests.
http://libguestfs.org/virt-v2v
#4 lore raw

PATCH 1/1 arm64: Fix text patching logic when using fixmap

In reply to Marc Zyngier
Quoted 51 lines · Marc Zyngier

Patch 2f896d586610 ("arm64: use fixmap for text patching") changed the way we patch the kernel text, using a fixmap when the kernel or modules are flagged as read only.

Unfortunately, a flaw in the logic makes it fall over when patching modules without CONFIG_DEBUG_SET_MODULE_RONX enabled:

[...]

[   32.032636] Call trace:
[   32.032716] [<fffffe00003da0dc>] __copy_to_user+0x2c/0x60
[   32.032837] [<fffffe0000099f08>] __aarch64_insn_write+0x94/0xf8
[   32.033027] [<fffffe000009a0a0>] aarch64_insn_patch_text_nosync+0x18/0x58
[   32.033200] [<fffffe000009c3ec>] ftrace_modify_code+0x58/0x84
[   32.033363] [<fffffe000009c4e4>] ftrace_make_nop+0x3c/0x58
[   32.033532] [<fffffe0000164420>] ftrace_process_locs+0x3d0/0x5c8
[   32.033709] [<fffffe00001661cc>] ftrace_module_init+0x28/0x34
[   32.033882] [<fffffe0000135148>] load_module+0xbb8/0xfc4
[   32.034044] [<fffffe0000135714>] SyS_finit_module+0x94/0xc4

[...]

This is triggered by the use of virt_to_page() on a module address, which ends to pointing to Nowhereland if you're lucky, or corrupt your precious data if not.

This patch fixes the logic by mimicking what is done on arm:
- If we're patching a module and CONFIG_DEBUG_SET_MODULE_RONX is set, use vmalloc_to_page().
- If we're patching the kernel and CONFIG_DEBUG_RODATA is set, use virt_to_page().
- Otherwise, use the provided address, as we can write to it directly.

Tested on 4.0-rc1 as a KVM guest.

Cc Kees Cook <[email protected]>
Cc Mark Rutland <[email protected]>
Cc Laura Abbott <[email protected]>
Cc Catalin Marinas <[email protected]>
Cc Will Deacon <[email protected]>
---
  arch/arm64/kernel/insn.c | 4 +++-
  1 file changed, 3 insertions(+), 1 deletion(-)
arch/arm64/kernel/insn.c +0 -0
index 27d4864..c8eca88 100644
@@ -87,8 +87,10 @@ static void __kprobes *patch_map(void *addr, int fixmap)
 

arch/arm64/kernel/insn.c static void __kprobes *patch_map(void *addr, int fixmap) · line 96

Quoted 10 lines
arch/arm64/kernel/insn.c +3 -1
  	if (module && IS_ENABLED(CONFIG_DEBUG_SET_MODULE_RONX))
  		page = vmalloc_to_page(addr);
-	else
+	else if (!module && IS_ENABLED(CONFIG_DEBUG_RODATA))
  		page = virt_to_page(addr);
+	else
+		return addr;
 
  	BUG_ON(!page);
  	set_fixmap(fixmap, page_to_phys(page));

Looks like I dropped this incorrectly between v6 and v7.

Signature · 3 lines
Qualcomm Innovation Center, Inc.
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project
#5 lore raw

PATCH 1/1 arm64: Fix text patching logic when using fixmap

In reply to Marc Zyngier
Quoted 33 lines · Marc Zyngier

Patch 2f896d586610 ("arm64: use fixmap for text patching") changed the way we patch the kernel text, using a fixmap when the kernel or modules are flagged as read only.

Unfortunately, a flaw in the logic makes it fall over when patching modules without CONFIG_DEBUG_SET_MODULE_RONX enabled:

[...]

[   32.032636] Call trace:
[   32.032716] [<fffffe00003da0dc>] __copy_to_user+0x2c/0x60
[   32.032837] [<fffffe0000099f08>] __aarch64_insn_write+0x94/0xf8
[   32.033027] [<fffffe000009a0a0>] aarch64_insn_patch_text_nosync+0x18/0x58
[   32.033200] [<fffffe000009c3ec>] ftrace_modify_code+0x58/0x84
[   32.033363] [<fffffe000009c4e4>] ftrace_make_nop+0x3c/0x58
[   32.033532] [<fffffe0000164420>] ftrace_process_locs+0x3d0/0x5c8
[   32.033709] [<fffffe00001661cc>] ftrace_module_init+0x28/0x34
[   32.033882] [<fffffe0000135148>] load_module+0xbb8/0xfc4
[   32.034044] [<fffffe0000135714>] SyS_finit_module+0x94/0xc4

[...]

This is triggered by the use of virt_to_page() on a module address, which ends to pointing to Nowhereland if you're lucky, or corrupt your precious data if not.

This patch fixes the logic by mimicking what is done on arm:
- If we're patching a module and CONFIG_DEBUG_SET_MODULE_RONX is set, use vmalloc_to_page().
- If we're patching the kernel and CONFIG_DEBUG_RODATA is set, use virt_to_page().
- Otherwise, use the provided address, as we can write to it directly.

Tested on 4.0-rc1 as a KVM guest.

Quoted 7 lines
Cc Kees Cook <[email protected]>
Cc Mark Rutland <[email protected]>
Cc Laura Abbott <[email protected]>
Cc Catalin Marinas <[email protected]>
Cc Will Deacon <[email protected]>

This fixes a crash on boot in the AHCI driver experienced by a test Fedora Hosted arm64 kernel when running on various platforms.

Tested by me on AMD Seattle and confirmed fixing the issue.

Quoted 21 lines
---
 arch/arm64/kernel/insn.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
arch/arm64/kernel/insn.c +3 -1
index 27d4864..c8eca88 100644
@@ -87,8 +87,10 @@ static void __kprobes *patch_map(void *addr, int fixmap)
 
 	if (module && IS_ENABLED(CONFIG_DEBUG_SET_MODULE_RONX))
 		page = vmalloc_to_page(addr);
-	else
+	else if (!module && IS_ENABLED(CONFIG_DEBUG_RODATA))
 		page = virt_to_page(addr);
+	else
+		return addr;
 
 	BUG_ON(!page);
 	set_fixmap(fixmap, page_to_phys(page));
#6 lore raw

PATCH 1/1 arm64: Fix text patching logic when using fixmap

Quoted 42 lines · Jon Masters
Quoted 33 lines · Marc Zyngier

Patch 2f896d586610 ("arm64: use fixmap for text patching") changed the way we patch the kernel text, using a fixmap when the kernel or modules are flagged as read only.

Unfortunately, a flaw in the logic makes it fall over when patching modules without CONFIG_DEBUG_SET_MODULE_RONX enabled:

[...]

[   32.032636] Call trace:
[   32.032716] [<fffffe00003da0dc>] __copy_to_user+0x2c/0x60
[   32.032837] [<fffffe0000099f08>] __aarch64_insn_write+0x94/0xf8
[   32.033027] [<fffffe000009a0a0>] aarch64_insn_patch_text_nosync+0x18/0x58
[   32.033200] [<fffffe000009c3ec>] ftrace_modify_code+0x58/0x84
[   32.033363] [<fffffe000009c4e4>] ftrace_make_nop+0x3c/0x58
[   32.033532] [<fffffe0000164420>] ftrace_process_locs+0x3d0/0x5c8
[   32.033709] [<fffffe00001661cc>] ftrace_module_init+0x28/0x34
[   32.033882] [<fffffe0000135148>] load_module+0xbb8/0xfc4
[   32.034044] [<fffffe0000135714>] SyS_finit_module+0x94/0xc4

[...]

This is triggered by the use of virt_to_page() on a module address, which ends to pointing to Nowhereland if you're lucky, or corrupt your precious data if not.

This patch fixes the logic by mimicking what is done on arm:
- If we're patching a module and CONFIG_DEBUG_SET_MODULE_RONX is set, use vmalloc_to_page().
- If we're patching the kernel and CONFIG_DEBUG_RODATA is set, use virt_to_page().
- Otherwise, use the provided address, as we can write to it directly.

Tested on 4.0-rc1 as a KVM guest.

Quoted 7 lines
Cc Kees Cook <[email protected]>
Cc Mark Rutland <[email protected]>
Cc Laura Abbott <[email protected]>
Cc Catalin Marinas <[email protected]>
Cc Will Deacon <[email protected]>
Quoted 6 lines

This fixes a crash on boot in the AHCI driver experienced by a test Fedora Hosted arm64 kernel when running on various platforms.

Tested by me on AMD Seattle and confirmed fixing the issue.

(In case anyone cares also tested on APM Mustang and it fixes the same)