x86/pkeys: add check for pkey "overflow"
Dave Hansen
maintainerPATCH 1/1 x86/pkeys: add check for pkey "overflow"
Alex Shi reported the pkey macros above arch_set_user_pkey_access() to be unused. They are unused, and even refer to a nonexistent CONFIG option.
But, they might have served a good use, which was to ensure that the code does not try to set values that would not fit in the PKRU register. As it stands, a too-large 'pkey' value would be likely to silently overflow the u32 new_pkru_bits.
Add a check to look for overflows. Also add a comment to remind any future developer to closely examine the types used to store pkey values if arch_max_pkey() ever changes.
This boots and passes the x86 pkey selftests.
---
b/arch/x86/include/asm/pkeys.h | 5 +++++ b/arch/x86/kernel/fpu/xstate.c | 9 +++++++-- 2 files changed, 12 insertions(+), 2 deletions(-)
arch/x86/kernel/fpu/xstate.c renamed from arch/x86/kernel/fpu/xstate.c~pkey-check-pkru-shift +7 -2
@@ -902,8 +902,6 @@ const void *get_xsave_field_ptr(int xfea#ifdef CONFIG_ARCH_HAS_PKEYS -#define NR_VALID_PKRU_BITS (CONFIG_NR_PROTECTION_KEYS * 2) -#define PKRU_VALID_MASK (NR_VALID_PKRU_BITS - 1)/* * This will go out and modify PKRU register to set the access * rights for @pkey to @init_val.
@@ -922,6 +920,13 @@ int arch_set_user_pkey_access(struct tasif (!boot_cpu_has (X86_FEATURE_OSPKE ))return -EINVAL ; +/* +* This code should only be called with valid 'pkey' +* values originating from in-kernel users. Complain +* if a bad value is observed. +*/ +WARN_ON_ONCE (pkey >=arch_max_pkey ()); +/* Set the bits we need in PKRU: */ if (init_val &PKEY_DISABLE_ACCESS ) new_pkru_bits |=PKRU_AD_BIT ;
arch/x86/include/asm/pkeys.h renamed from arch/x86/include/asm/pkeys.h~pkey-check-pkru-shift +5 -0
@@ -4,6 +4,11 @@#define ARCH_DEFAULT_PKEY 0 +/* +* If more than 16 keys are ever supported, a thorough audit +* will be necessary to ensure that the types that store key +* numbers and masks have sufficient capacity. +*/ #define arch_max_pkey () (boot_cpu_has(X86_FEATURE_OSPKE) ? 16 : 1) extern int arch_set_user_pkey_access(struct task_struct *tsk, int pkey,
_
Quoted 6 lines · Dave Hansen
Alex Shi reported the pkey macros above
arch_set_user_pkey_access()to be unused. They are unused, and even refer to a nonexistent CONFIG option.
arch/x86/kernel/fpu/xstate.c int arch_set_user_pkey_access(struct tas · line 928
Quoted 9 lines
arch/x86/kernel/fpu/xstate.c renamed from arch/x86/kernel/fpu/xstate.c~pkey-check-pkru-shift +6
-0@@ -922,6 +920,13 @@ int arch_set_user_pkey_access(struct tasif (!boot_cpu_has (X86_FEATURE_OSPKE ))return -EINVAL ; +/* +* This code should only be called with valid 'pkey' +* values originating from in-kernel users. Complain +* if a bad value is observed. +*/ +WARN_ON_ONCE (pkey >=arch_max_pkey ());
Should not we rather abort this operation and exit with EINVAL or something similar instead of calling wrmsr with overflowed value? IOW,
if (pkey >= arch_max_pkey()) {
WARN_ON_ONCE(1);
return -EINVAL;
}Quoted 10 lines · Cyrill Gorcunov
arch/x86/kernel/fpu/xstate.c int arch_set_user_pkey_access(struct tas · line 928
Quoted 6 lines
arch/x86/kernel/fpu/xstate.c renamed from arch/x86/kernel/fpu/xstate.c~pkey-check-pkru-shift +6
-0+/* +* This code should only be called with valid 'pkey' +* values originating from in-kernel users. Complain +* if a bad value is observed. +*/ +WARN_ON_ONCE (pkey >=arch_max_pkey ());Should not we rather abort this operation and exit with EINVAL or something similar instead of calling wrmsr with overflowed value? IOW,
Quoted 4 lines
if (pkey >= arch_max_pkey()) {
WARN_ON_ONCE(1);
return -EINVAL;
}I don't feel strongly about it. The reason I didn't do that is to minimize the chance that this would cause any functional regression.
It's not a huge chance, but I've certainly fat-fingered my share of off-by-one bugs.
Quoted 12 lines · Dave Hansen
Quoted 11 lines · Cyrill Gorcunov
arch/x86/kernel/fpu/xstate.c int arch_set_user_pkey_access(struct tas · line 928
Quoted 6 lines
arch/x86/kernel/fpu/xstate.c renamed from arch/x86/kernel/fpu/xstate.c~pkey-check-pkru-shift +6
-0+/* +* This code should only be called with valid 'pkey' +* values originating from in-kernel users. Complain +* if a bad value is observed. +*/ +WARN_ON_ONCE (pkey >=arch_max_pkey ());Should not we rather abort this operation and exit with EINVAL or something similar instead of calling wrmsr with overflowed value? IOW,
Quoted 7 lines
Quoted 4 lines
if (pkey >= arch_max_pkey()) { WARN_ON_ONCE(1); return -EINVAL; }I don't feel strongly about it. The reason I didn't do that is to minimize the chance that this would cause any functional regression.
OK, I don't mind leaving just WARN_ON_ONCE.
Quoted 3 lines
It's not a huge chance, but I've certainly fat-fingered my share of off-by-one bugs.
Heh :)
[tip: x86/fpu] x86/pkeys: Add check for pkey "overflow"
In reply to Dave HansenThe following commit has been merged into the x86/fpu branch of tip:
The following commit has been merged into the x86/fpu branch of tip:
Commit-ID: 16171bffc829272d5e6014bad48f680cb50943d9 Gitweb: https://git.kernel.org/tip/16171bffc829272d5e6014bad48f680cb50943d9 Author: Dave Hansen <[email protected]> AuthorDate: Wed, 22 Jan 2020 08:53:46 -08:00 Committer: Borislav Petkov <[email protected]> CommitterDate: Mon, 24 Feb 2020 20:25:21 +01:00
x86/pkeys: Add check for pkey "overflow"
Alex Shi reported the pkey macros above arch_set_user_pkey_access() to be unused. They are unused, and even refer to a nonexistent CONFIG option.
But, they might have served a good use, which was to ensure that the code does not try to set values that would not fit in the PKRU register. As it stands, a too-large 'pkey' value would be likely to silently overflow the u32 new_pkru_bits.
Add a check to look for overflows. Also add a comment to remind any future developer to closely examine the types used to store pkey values if arch_max_pkey() ever changes.
This boots and passes the x86 pkey selftests.
--- arch/x86/include/asm/pkeys.h | 5 +++++ arch/x86/kernel/fpu/xstate.c | 9 +++++++-- 2 files changed, 12 insertions(+), 2 deletions(-)
arch/x86/include/asm/pkeys.h +5 -0
@@ -4,6 +4,11 @@#define ARCH_DEFAULT_PKEY 0 +/* +* If more than 16 keys are ever supported, a thorough audit +* will be necessary to ensure that the types that store key +* numbers and masks have sufficient capacity. +*/ #define arch_max_pkey () (boot_cpu_has(X86_FEATURE_OSPKE) ? 16 : 1) extern int arch_set_user_pkey_access(struct task_struct *tsk, int pkey,
arch/x86/kernel/fpu/xstate.c +7 -2
@@ -895,8 +895,6 @@ const void *get_xsave_field_ptr(int xfeature_nr)#ifdef CONFIG_ARCH_HAS_PKEYS -#define NR_VALID_PKRU_BITS (CONFIG_NR_PROTECTION_KEYS * 2) -#define PKRU_VALID_MASK (NR_VALID_PKRU_BITS - 1)/* * This will go out and modify PKRU register to set the access * rights for @pkey to @init_val.
@@ -915,6 +913,13 @@ int arch_set_user_pkey_access(struct task_struct *tsk, int pkey,if (!boot_cpu_has (X86_FEATURE_OSPKE ))return -EINVAL ; +/* +* This code should only be called with valid 'pkey' +* values originating from in-kernel users. Complain +* if a bad value is observed. +*/ +WARN_ON_ONCE (pkey >=arch_max_pkey ()); +/* Set the bits we need in PKRU: */ if (init_val &PKEY_DISABLE_ACCESS ) new_pkru_bits |=PKRU_AD_BIT ;