[PATCH v2] arm64: kpti: Fix the interaction between ASID switching and software PAN

On 16 January 2018 at 20:55, Ard Biesheuvel [off-list ref] wrote:

On 16 January 2018 at 17:35, Catalin Marinas [off-list ref] wrote:

quoted

On Tue, Jan 16, 2018 at 04:23:58PM +0000, Ard Biesheuvel wrote:

quoted

On 16 January 2018 at 12:11, James Morse [off-list ref] wrote:

quoted

On 15/01/18 17:58, Catalin Marinas wrote:

quoted

With ARM64_SW_TTBR0_PAN enabled, the exception entry code checks the
active ASID to decide whether user access was enabled (non-zero ASID)
when the exception was taken. On return from exception, if user access
was previously disabled, it re-instates TTBR0_EL1 from the per-thread
saved value (updated in switch_mm() or efi_set_pgd()).

Commit 7655abb95386 ("arm64: mm: Move ASID from TTBR0 to TTBR1") makes a
TTBR0_EL1 + ASID switching non-atomic. Subsequently, commit 27a921e75711
("arm64: mm: Fix and re-enable ARM64_SW_TTBR0_PAN") changes the
__uaccess_ttbr0_disable() function and asm macro to first write the
reserved TTBR0_EL1 followed by the ASID=0 update in TTBR1_EL1. If an
exception occurs between these two, the exception return code will
re-instate a valid TTBR0_EL1. Similar scenario can happen in
cpu_switch_mm() between setting the reserved TTBR0_EL1 and the ASID
update in cpu_do_switch_mm().

This patch reverts the entry.S check for ASID == 0 to TTBR0_EL1 and
disables the interrupts around the TTBR0_EL1 and ASID switching code in
__uaccess_ttbr0_disable(). It also ensures that, when returning from the
EFI runtime services, efi_set_pgd() doesn't leave a non-zero ASID in
TTBR1_EL1 by using uaccess_ttbr0_{enable,disable}.

The accesses to current_thread_info()->ttbr0 are updated to use
READ_ONCE/WRITE_ONCE.

As a safety measure, __uaccess_ttbr0_enable() always masks out any
existing non-zero ASID TTBR1_EL1 before writing in the new ASID.

Fixes: 27a921e75711 ("arm64: mm: Fix and re-enable ARM64_SW_TTBR0_PAN")
Acked-by: Will Deacon <redacted>
Reported-by: Ard Biesheuvel <redacted>
Cc: Ard Biesheuvel <redacted>
Reviewed-by: James Morse <james.morse@arm.com>
Cc: James Morse <james.morse@arm.com>
Co-developed-by: Marc Zyngier <redacted>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
---

Changes since v1 (no functional change):

- Modified efi_set_pgd() to pair uaccess_ttbr0_disable/enable
- Accesses to current_thread_info()->ttbr0 are passed through READ/WRITE_ONCE

I dropped James' tested-by, just in case I messed something up.

Tested as before:
Tested-by: James Morse <james.morse@arm.com>

Works for me as well (Gnome Shell desktop + Chromium browser)

Tested-by: Ard Biesheuvel <redacted>

Great. Thanks both for re-testing. I'll push the patch out to
for-next/core.

Unfortunately, I just got another

make: ../sysdeps/nptl/fork.c:156: __libc_fork: Assertion
`THREAD_GETMEM (self, tid) != ppid' failed.

while compiling a kernel, which is the exact same symptom I had before.

Unfortunately, it appears I am incompetent. It seems the kernel I
tested did not have this v2 applied (nor any other version of this
patch)

I am rebuilding now, I will report back today if I still run into anything.

Apologies for the noise.