Content-Type: text/plain; charset=utf-8 Mime-Version: 1.0 (Mac OS X Mail 12.4 \(3445.104.8\)) Subject: Re: [PATCH v4 03/23] x86/mm: Introduce temporary mm structs From: Nadav Amit In-Reply-To: <20190425162620.GA5199@zn.tnic> Date: Thu, 25 Apr 2019 10:37:35 -0700 Content-Transfer-Encoding: quoted-printable Message-Id: References: <20190422185805.1169-1-rick.p.edgecombe@intel.com> <20190422185805.1169-4-rick.p.edgecombe@intel.com> <20190425162620.GA5199@zn.tnic> To: Andy Lutomirski Cc: Rick Edgecombe , Ingo Molnar , LKML , X86 ML , "H. Peter Anvin" , Thomas Gleixner , Dave Hansen , Peter Zijlstra , Damian Tometzki , linux-integrity , LSM List , Andrew Morton , Kernel Hardening , Linux-MM , Will Deacon , Ard Biesheuvel , Kristen Carlson Accardi , "Dock, Deneen T" , Kees Cook , Dave Hansen , Borislav Petkov List-ID: > On Apr 25, 2019, at 9:26 AM, Borislav Petkov wrote: >=20 > On Mon, Apr 22, 2019 at 11:57:45AM -0700, Rick Edgecombe wrote: >> From: Andy Lutomirski >>=20 >> Using a dedicated page-table for temporary PTEs prevents other cores >> from using - even speculatively - these PTEs, thereby providing two >> benefits: >>=20 >> (1) Security hardening: an attacker that gains kernel memory writing >> abilities cannot easily overwrite sensitive data. >>=20 >> (2) Avoiding TLB shootdowns: the PTEs do not need to be flushed in >> remote page-tables. >>=20 >> To do so a temporary mm_struct can be used. Mappings which are = private >> for this mm can be set in the userspace part of the address-space. >> During the whole time in which the temporary mm is loaded, interrupts >> must be disabled. >>=20 >> The first use-case for temporary mm struct, which will follow, is for >> poking the kernel text. >>=20 >> [ Commit message was written by Nadav Amit ] >>=20 >> Cc: Kees Cook >> Cc: Dave Hansen >> Acked-by: Peter Zijlstra (Intel) >> Reviewed-by: Masami Hiramatsu >> Tested-by: Masami Hiramatsu >> Signed-off-by: Andy Lutomirski >> Signed-off-by: Nadav Amit >> Signed-off-by: Rick Edgecombe >> --- >> arch/x86/include/asm/mmu_context.h | 33 = ++++++++++++++++++++++++++++++ >> 1 file changed, 33 insertions(+) >>=20 >> diff --git a/arch/x86/include/asm/mmu_context.h = b/arch/x86/include/asm/mmu_context.h >> index 19d18fae6ec6..d684b954f3c0 100644 >> --- a/arch/x86/include/asm/mmu_context.h >> +++ b/arch/x86/include/asm/mmu_context.h >> @@ -356,4 +356,37 @@ static inline unsigned long = __get_current_cr3_fast(void) >> return cr3; >> } >>=20 >> +typedef struct { >> + struct mm_struct *prev; >> +} temp_mm_state_t; >> + >> +/* >> + * Using a temporary mm allows to set temporary mappings that are = not accessible >> + * by other cores. Such mappings are needed to perform sensitive = memory writes >=20 > s/cores/CPUs/g >=20 > Yeah, the concept of a thread of execution we call a CPU in the = kernel, > I'd say. No matter if it is one of the hyperthreads or a single thread > in core. >=20 >> + * that override the kernel memory protections (e.g., W^X), without = exposing the >> + * temporary page-table mappings that are required for these write = operations to >> + * other cores. >=20 > Ditto. >=20 >> Using temporary mm also allows to avoid TLB shootdowns when the >=20 > Using a .. >=20 >> + * mapping is torn down. >> + * >=20 > Nice commenting. >=20 >> + * Context: The temporary mm needs to be used exclusively by a = single core. To >> + * harden security IRQs must be disabled while the = temporary mm is > ^ > , >=20 >> + * loaded, thereby preventing interrupt handler bugs from = overriding >> + * the kernel memory protection. >> + */ >> +static inline temp_mm_state_t use_temporary_mm(struct mm_struct *mm) >> +{ >> + temp_mm_state_t state; >> + >> + lockdep_assert_irqs_disabled(); >> + state.prev =3D this_cpu_read(cpu_tlbstate.loaded_mm); >> + switch_mm_irqs_off(NULL, mm, current); >> + return state; >> +} >> + >> +static inline void unuse_temporary_mm(temp_mm_state_t prev) >> +{ >> + lockdep_assert_irqs_disabled(); >> + switch_mm_irqs_off(NULL, prev.prev, current); >=20 > I think this code would be more readable if you call that > temp_mm_state_t variable "temp_state" and the mm_struct pointer "mm" = and > then you have: >=20 > switch_mm_irqs_off(NULL, temp_state.mm, current); >=20 > And above you'll have: >=20 > temp_state.mm =3D ... Andy, please let me know whether you are fine with this change and = I=E2=80=99ll incorporate it.=