From: Yang Shi <yang@os.amperecomputing.com>
To: Catalin Marinas <catalin.marinas@arm.com>
Cc: will@kernel.org, anshuman.khandual@arm.com,
scott@os.amperecomputing.com, cl@gentwo.org,
linux-arm-kernel@lists.infradead.org,
linux-kernel@vger.kernel.org
Subject: Re: [v3 PATCH] arm64: mm: force write fault for atomic RMW instructions
Date: Wed, 5 Jun 2024 10:31:51 -0700 [thread overview]
Message-ID: <ee4d5664-a426-48d3-ad40-b33ac5f44a7f@os.amperecomputing.com> (raw)
In-Reply-To: <ZmCYLvb_JlkCuc_u@arm.com>
>> @@ -557,6 +587,11 @@ static int __kprobes do_page_fault(unsigned long far, unsigned long esr,
>> /* It was write fault */
>> vm_flags = VM_WRITE;
>> mm_flags |= FAULT_FLAG_WRITE;
>> + } else if (is_el0_atomic_instr(regs)) {
>> + /* Force write fault */
>> + vm_flags = VM_WRITE;
>> + mm_flags |= FAULT_FLAG_WRITE;
>> + force_write = true;
>> } else {
>> /* It was read fault */
>> vm_flags = VM_READ;
>> @@ -586,6 +621,14 @@ static int __kprobes do_page_fault(unsigned long far, unsigned long esr,
>> if (!vma)
>> goto lock_mmap;
>>
>> + /* vma flags don't allow write, undo force write */
>> + if (force_write && !(vma->vm_flags & VM_WRITE)) {
>> + vm_flags |= VM_READ;
>> + if (!alternative_has_cap_unlikely(ARM64_HAS_EPAN))
>> + vm_flags |= VM_EXEC;
>> + mm_flags &= ~FAULT_FLAG_WRITE;
>> + }
> Ah, this revert to the non-write flags doesn't look great as we
> basically duplicate the 'else' block in the original check. So, it
> probably look better as per your earlier patch to just do the
> instruction read just before the !(vma->vm_flags & flags) check,
> something like:
>
> if ((vma->vm_flags & VM_WRITE) && is_el0_atomic_instr(regs)) {
> vm_flags = VM_WRITE;
> mm_flags |= FAULT_FLAG_WRITE;
> }
>
> This way we also only read the instruction if the vma is writeable. I
> think it's fine to do this under the vma lock since we have
> pagefault_disable() for the insn read.
I think we also need to skip this for write fault and instruction fault.
Some something like:
@@ -529,6 +557,7 @@ static int __kprobes do_page_fault(unsigned long
far, unsigned long esr,
unsigned int mm_flags = FAULT_FLAG_DEFAULT;
unsigned long addr = untagged_addr(far);
struct vm_area_struct *vma;
+ bool may_force_write = false;
if (kprobe_page_fault(regs, esr))
return 0;
@@ -565,6 +594,7 @@ static int __kprobes do_page_fault(unsigned long
far, unsigned long esr,
/* If EPAN is absent then exec implies read */
if (!alternative_has_cap_unlikely(ARM64_HAS_EPAN))
vm_flags |= VM_EXEC;
+ may_force_write = true;
}
if (is_ttbr0_addr(addr) && is_el1_permission_fault(addr, esr,
regs)) {
@@ -586,6 +616,12 @@ static int __kprobes do_page_fault(unsigned long
far, unsigned long esr,
if (!vma)
goto lock_mmap;
+ if (may_force_write && (vma->vm_flags & VM_WRITE) &&
+ is_el0_atomic_instr(regs)) {
+ vm_flags = VM_WRITE;
+ mm_flags |= FAULT_FLAG_WRITE;
+ }
+
if (!(vma->vm_flags & vm_flags)) {
vma_end_read(vma);
goto lock_mmap;
>
prev parent reply other threads:[~2024-06-05 17:31 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-06-04 17:15 [v3 PATCH] arm64: mm: force write fault for atomic RMW instructions Yang Shi
2024-06-05 16:54 ` Catalin Marinas
2024-06-05 17:12 ` Yang Shi
2024-06-05 17:31 ` Yang Shi [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=ee4d5664-a426-48d3-ad40-b33ac5f44a7f@os.amperecomputing.com \
--to=yang@os.amperecomputing.com \
--cc=anshuman.khandual@arm.com \
--cc=catalin.marinas@arm.com \
--cc=cl@gentwo.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=scott@os.amperecomputing.com \
--cc=will@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox