dept needs to notice every entrance from user to kernel mode to treat
every kernel context independently when tracking wait-event dependencies.
Roughly, system call and user oriented fault are the cases.

Make dept aware of the entrances of arm64 and add support
CONFIG_ARCH_HAS_DEPT_SUPPORT to arm64.

Signed-off-by: Byungchul Park <byungchul@sk.com>
---
 arch/arm64/Kconfig          | 1 +
 arch/arm64/kernel/syscall.c | 7 +++++++
 arch/arm64/mm/fault.c       | 7 +++++++
 3 files changed, 15 insertions(+)

diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
index e9bbfacc35a6..a8fab2c052dc 100644
--- a/arch/arm64/Kconfig
+++ b/arch/arm64/Kconfig
@@ -281,6 +281,7 @@ config ARM64
 	select USER_STACKTRACE_SUPPORT
 	select VDSO_GETRANDOM
 	select VMAP_STACK
+	select ARCH_HAS_DEPT_SUPPORT
 	help
 	  ARM 64-bit (AArch64) Linux support.
 
diff --git a/arch/arm64/kernel/syscall.c b/arch/arm64/kernel/syscall.c
index c442fcec6b9e..bbd306335179 100644
--- a/arch/arm64/kernel/syscall.c
+++ b/arch/arm64/kernel/syscall.c
@@ -7,6 +7,7 @@
 #include <linux/ptrace.h>
 #include <linux/randomize_kstack.h>
 #include <linux/syscalls.h>
+#include <linux/dept.h>
 
 #include <asm/debug-monitors.h>
 #include <asm/exception.h>
@@ -96,6 +97,12 @@ static void el0_svc_common(struct pt_regs *regs, int scno, int sc_nr,
 	 * (Similarly for HVC and SMC elsewhere.)
 	 */
 
+	/*
+	 * This is a system call from user mode.  Make dept work with a
+	 * new kernel mode context.
+	 */
+	dept_update_cxt();
+
 	if (flags & _TIF_MTE_ASYNC_FAULT) {
 		/*
 		 * Process the asynchronous tag check fault before the actual
diff --git a/arch/arm64/mm/fault.c b/arch/arm64/mm/fault.c
index d816ff44faff..96827b999d18 100644
--- a/arch/arm64/mm/fault.c
+++ b/arch/arm64/mm/fault.c
@@ -26,6 +26,7 @@
 #include <linux/pkeys.h>
 #include <linux/preempt.h>
 #include <linux/hugetlb.h>
+#include <linux/dept.h>
 
 #include <asm/acpi.h>
 #include <asm/bug.h>
@@ -622,6 +623,12 @@ static int __kprobes do_page_fault(unsigned long far, unsigned long esr,
 	if (!(mm_flags & FAULT_FLAG_USER))
 		goto lock_mmap;
 
+	/*
+	 * This fault comes from user mode.  Make dept work with a new
+	 * kernel mode context.
+	 */
+	dept_update_cxt();
+
 	vma = lock_vma_under_rcu(mm, addr);
 	if (!vma)
 		goto lock_mmap;
-- 
2.17.1

On Thu, Oct 02, 2025 at 05:12:09PM +0900, Byungchul Park wrote:
> dept needs to notice every entrance from user to kernel mode to treat
> every kernel context independently when tracking wait-event dependencies.
> Roughly, system call and user oriented fault are the cases.
> 
> Make dept aware of the entrances of arm64 and add support
> CONFIG_ARCH_HAS_DEPT_SUPPORT to arm64.
> 
> Signed-off-by: Byungchul Park <byungchul@sk.com>
> ---
>  arch/arm64/Kconfig          | 1 +
>  arch/arm64/kernel/syscall.c | 7 +++++++
>  arch/arm64/mm/fault.c       | 7 +++++++
>  3 files changed, 15 insertions(+)
> 
> diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
> index e9bbfacc35a6..a8fab2c052dc 100644
> --- a/arch/arm64/Kconfig
> +++ b/arch/arm64/Kconfig
> @@ -281,6 +281,7 @@ config ARM64
>  	select USER_STACKTRACE_SUPPORT
>  	select VDSO_GETRANDOM
>  	select VMAP_STACK
> +	select ARCH_HAS_DEPT_SUPPORT
>  	help
>  	  ARM 64-bit (AArch64) Linux support.
>  
> diff --git a/arch/arm64/kernel/syscall.c b/arch/arm64/kernel/syscall.c
> index c442fcec6b9e..bbd306335179 100644
> --- a/arch/arm64/kernel/syscall.c
> +++ b/arch/arm64/kernel/syscall.c
> @@ -7,6 +7,7 @@
>  #include <linux/ptrace.h>
>  #include <linux/randomize_kstack.h>
>  #include <linux/syscalls.h>
> +#include <linux/dept.h>
>  
>  #include <asm/debug-monitors.h>
>  #include <asm/exception.h>
> @@ -96,6 +97,12 @@ static void el0_svc_common(struct pt_regs *regs, int scno, int sc_nr,
>  	 * (Similarly for HVC and SMC elsewhere.)
>  	 */
>  
> +	/*
> +	 * This is a system call from user mode.  Make dept work with a
> +	 * new kernel mode context.
> +	 */
> +	dept_update_cxt();

As Mark Brown pointed out in his replies, this patch is missing a whole
bunch of cases and does not work correctly as-is.

As Dave Hansen pointed out on the x86 patch, you shouldn't do this
piecemeal in architecture code, and should instead work with the
existing context tracking, e.g. by adding logic to
enter_from_user_mode() and exit_to_user_mode(), or by reusing some
existing context tracking logic that's called there.

Mark.

On Fri, Oct 03, 2025 at 03:36:42PM +0100, Mark Rutland wrote:
> On Thu, Oct 02, 2025 at 05:12:09PM +0900, Byungchul Park wrote:
> > dept needs to notice every entrance from user to kernel mode to treat
> > every kernel context independently when tracking wait-event dependencies.
> > Roughly, system call and user oriented fault are the cases.
> >
> > Make dept aware of the entrances of arm64 and add support
> > CONFIG_ARCH_HAS_DEPT_SUPPORT to arm64.
> >
> > Signed-off-by: Byungchul Park <byungchul@sk.com>
> > ---
> >  arch/arm64/Kconfig          | 1 +
> >  arch/arm64/kernel/syscall.c | 7 +++++++
> >  arch/arm64/mm/fault.c       | 7 +++++++
> >  3 files changed, 15 insertions(+)
> >
> > diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
> > index e9bbfacc35a6..a8fab2c052dc 100644
> > --- a/arch/arm64/Kconfig
> > +++ b/arch/arm64/Kconfig
> > @@ -281,6 +281,7 @@ config ARM64
> >       select USER_STACKTRACE_SUPPORT
> >       select VDSO_GETRANDOM
> >       select VMAP_STACK
> > +     select ARCH_HAS_DEPT_SUPPORT
> >       help
> >         ARM 64-bit (AArch64) Linux support.
> >
> > diff --git a/arch/arm64/kernel/syscall.c b/arch/arm64/kernel/syscall.c
> > index c442fcec6b9e..bbd306335179 100644
> > --- a/arch/arm64/kernel/syscall.c
> > +++ b/arch/arm64/kernel/syscall.c
> > @@ -7,6 +7,7 @@
> >  #include <linux/ptrace.h>
> >  #include <linux/randomize_kstack.h>
> >  #include <linux/syscalls.h>
> > +#include <linux/dept.h>
> >
> >  #include <asm/debug-monitors.h>
> >  #include <asm/exception.h>
> > @@ -96,6 +97,12 @@ static void el0_svc_common(struct pt_regs *regs, int scno, int sc_nr,
> >        * (Similarly for HVC and SMC elsewhere.)
> >        */
> >
> > +     /*
> > +      * This is a system call from user mode.  Make dept work with a
> > +      * new kernel mode context.
> > +      */
> > +     dept_update_cxt();
> 
> As Mark Brown pointed out in his replies, this patch is missing a whole
> bunch of cases and does not work correctly as-is.
> 
> As Dave Hansen pointed out on the x86 patch, you shouldn't do this
> piecemeal in architecture code, and should instead work with the
> existing context tracking, e.g. by adding logic to
> enter_from_user_mode() and exit_to_user_mode(), or by reusing some

I will consider it.  However, I need to check if there are not any waits
and events before enter_from_user_mode(), or after exit_to_user_mode()
since those functions aren't the outmost functions for kernel mode C
code anyway.

	Byungchul

> existing context tracking logic that's called there.
> 
> Mark.

On Thu, Oct 02, 2025 at 05:12:09PM +0900, Byungchul Park wrote:
> dept needs to notice every entrance from user to kernel mode to treat
> every kernel context independently when tracking wait-event dependencies.
> Roughly, system call and user oriented fault are the cases.
> 
> Make dept aware of the entrances of arm64 and add support
> CONFIG_ARCH_HAS_DEPT_SUPPORT to arm64.

The description of what needs to be tracked probably needs some
tightening up here, it's not clear to me for example why exceptions for
mops or the vector extensions aren't included here, or what the
distinction is with error faults like BTI or GCS not being tracked?

On Thu, Oct 02, 2025 at 12:39:31PM +0100, Mark Brown wrote:
> On Thu, Oct 02, 2025 at 05:12:09PM +0900, Byungchul Park wrote:
> > dept needs to notice every entrance from user to kernel mode to treat
> > every kernel context independently when tracking wait-event dependencies.
> > Roughly, system call and user oriented fault are the cases.
> > 
> > Make dept aware of the entrances of arm64 and add support
> > CONFIG_ARCH_HAS_DEPT_SUPPORT to arm64.
> 
> The description of what needs to be tracked probably needs some
> tightening up here, it's not clear to me for example why exceptions for
> mops or the vector extensions aren't included here, or what the
> distinction is with error faults like BTI or GCS not being tracked?

Thanks for the feedback but I'm afraid I don't get you.  Can you explain
in more detail with example?

JFYI, pairs of wait and its event need to be tracked to see if each
event can be prevented from being reachable by other waits like:

   context X				context Y

   lock L
   ...
   initiate event A context		start toward event A
   ...					...
   wait A // wait for event A and	lock L // wait for unlock L and
          // prevent unlock L		       // prevent event A
   ...					...
   unlock L				unlock L
					...
					event A

I meant things like this need to be tracked.

	Byungchul

On Fri, Oct 03, 2025 at 10:46:41AM +0900, Byungchul Park wrote:
> On Thu, Oct 02, 2025 at 12:39:31PM +0100, Mark Brown wrote:
> > On Thu, Oct 02, 2025 at 05:12:09PM +0900, Byungchul Park wrote:
> > > dept needs to notice every entrance from user to kernel mode to treat
> > > every kernel context independently when tracking wait-event dependencies.
> > > Roughly, system call and user oriented fault are the cases.

> > > Make dept aware of the entrances of arm64 and add support
> > > CONFIG_ARCH_HAS_DEPT_SUPPORT to arm64.

> > The description of what needs to be tracked probably needs some
> > tightening up here, it's not clear to me for example why exceptions for
> > mops or the vector extensions aren't included here, or what the
> > distinction is with error faults like BTI or GCS not being tracked?

> Thanks for the feedback but I'm afraid I don't get you.  Can you explain
> in more detail with example?

Your commit log says we need to track every entrance from user mode to
kernel mode but the code only adds tracking to syscalls and some memory
faults.  The exception types listed above (and some others) also result
in entries to the kernel from userspace.

> JFYI, pairs of wait and its event need to be tracked to see if each
> event can be prevented from being reachable by other waits like:

>    context X				context Y
> 
>    lock L
>    ...
>    initiate event A context		start toward event A
>    ...					...
>    wait A // wait for event A and	lock L // wait for unlock L and
>           // prevent unlock L		       // prevent event A
>    ...					...
>    unlock L				unlock L
> 					...
> 					event A

> I meant things like this need to be tracked.

I don't think that's at all clear from the above context, and the
handling for some of the above exception types (eg, the vector
extensions) includes taking locks.

On Fri, Oct 03, 2025 at 12:33:03PM +0100, Mark Brown wrote:
> On Fri, Oct 03, 2025 at 10:46:41AM +0900, Byungchul Park wrote:
> > On Thu, Oct 02, 2025 at 12:39:31PM +0100, Mark Brown wrote:
> > > On Thu, Oct 02, 2025 at 05:12:09PM +0900, Byungchul Park wrote:
> > > > dept needs to notice every entrance from user to kernel mode to treat
> > > > every kernel context independently when tracking wait-event dependencies.
> > > > Roughly, system call and user oriented fault are the cases.
> 
> > > > Make dept aware of the entrances of arm64 and add support
> > > > CONFIG_ARCH_HAS_DEPT_SUPPORT to arm64.
> 
> > > The description of what needs to be tracked probably needs some
> > > tightening up here, it's not clear to me for example why exceptions for
> > > mops or the vector extensions aren't included here, or what the
> > > distinction is with error faults like BTI or GCS not being tracked?
> 
> > Thanks for the feedback but I'm afraid I don't get you.  Can you explain
> > in more detail with example?
> 
> Your commit log says we need to track every entrance from user mode to
> kernel mode but the code only adds tracking to syscalls and some memory
> faults.  The exception types listed above (and some others) also result
> in entries to the kernel from userspace.

You're right.  Each kernel mode context needs to be tracked
independently.  Just for your information, context ID is used for making
DEPT only track waits and events within each context, preventing
tracking those across independent contexts to avoid false positives.

Currently, irq, fault, and system calls are covered.  It should be taken
into account if any other exceptions can include waits and events anyway.

> > JFYI, pairs of wait and its event need to be tracked to see if each
> > event can be prevented from being reachable by other waits like:
> 
> >    context X				context Y
> > 
> >    lock L
> >    ...
> >    initiate event A context		start toward event A
> >    ...					...
> >    wait A // wait for event A and	lock L // wait for unlock L and
> >           // prevent unlock L		       // prevent event A
> >    ...					...
> >    unlock L				unlock L
> > 					...
> > 					event A
> 
> > I meant things like this need to be tracked.
> 
> I don't think that's at all clear from the above context, and the
> handling for some of the above exception types (eg, the vector
> extensions) includes taking locks.

A trivial thing to mention, each typical lock pair, lock and unlock, can
only happen within each independent context, not across different
contexts.  So the context ID is not necessary for typical locks.

   exception X

   lock A;
   ...
   unlock A; // already guarateed to unlock A in the context that lock A
             // has been acqured in.
   ...
   finish exception X and return back to user mode;

But yes, as you concern, we should take into account all the exceptions
if any can include general waits and events in it, to avoid unnecessary
false positives.

Thank you!

	Byungchul