[PATCH] arm64: Call debug_traps_init() from trap_init() to help early kgdb

[PATCH] arm64: Call debug_traps_init() from trap_init() to help early kgdb

[Douglas Anderson]

A new kgdb feature will soon land (kgdb_earlycon) that lets us run
kgdb much earlier.  In order for everything to work properly it's
important that the break hook is setup by the time we process
"kgdbwait".

Right now the break hook is setup in debug_traps_init() and that's
called from arch_initcall().  That's a bit too late since
kgdb_earlycon really needs things to be setup by the time the system
calls dbg_late_init().

We could fix this by adding call_break_hook() into early_brk64() and
that works fine.  However, it's a little ugly.  Instead, let's just
add a call to debug_traps_init() straight from trap_init().  There's
already a documented dependency between trap_init() and
debug_traps_init() and this makes the dependency more obvious rather
than just relying on a comment.

NOTE: this solution isn't early enough to let us select the
"ARCH_HAS_EARLY_DEBUG" KConfig option that is introduced by the
kgdb_earlycon patch series.  That would only be set if we could do
breakpoints when early params are parsed.  This patch only enables
"late early" breakpoints, AKA breakpoints when dbg_late_init() is
called.  It's expected that this should be fine for most people.

It should also be noted that if you crash you can still end up in kgdb
earlier than debug_traps_init().  Since you don't need breakpoints to
debug a crash that's fine.

Suggested-by: Will Deacon <will@kernel.org>
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will@kernel.org>
---
This replaces the patch ("arm64: Add call_break_hook() to
early_brk64() for early kgdb") in my recent kgdb series [1].  If I end
up re-posting that series again I'll include this patch as a
replacement, but I'm sending it separately to avoid spamming a pile of
people another time with a 12-patch series.

Note that, because it doesn't select the "ARCH_HAS_EARLY_DEBUG"
KConfig option it could be landed standalone.  However, it's still
probably better to land together with that patch series.

If the kgdb_earlycon patch series lands without this patch then
kgdbwait + kgdb_earlycon won't work well on arm64, but there would be
no other bad side effects.

If this patch lands without the kgdb_earlycon patch series then there
will be no known problems.

[1] https://lore.kernel.org/r/20200507130644.v4.5.I22067ad43e77ddfd4b64c2d49030628480f9e8d9@changeid

 arch/arm64/include/asm/debug-monitors.h | 2 ++
 arch/arm64/kernel/debug-monitors.c      | 4 +---
 arch/arm64/kernel/traps.c               | 2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/arch/arm64/include/asm/debug-monitors.h b/arch/arm64/include/asm/debug-monitors.h
index 7619f473155f..e5ceea213e39 100644
--- a/arch/arm64/include/asm/debug-monitors.h
+++ b/arch/arm64/include/asm/debug-monitors.h
@@ -125,5 +125,7 @@ static inline int reinstall_suspended_bps(struct pt_regs *regs)
 
 int aarch32_break_handler(struct pt_regs *regs);
 
+void debug_traps_init(void);
+
 #endif	/* __ASSEMBLY */
 #endif	/* __ASM_DEBUG_MONITORS_H */
diff --git a/arch/arm64/kernel/debug-monitors.c b/arch/arm64/kernel/debug-monitors.c
index 48222a4760c2..15e80c876d46 100644
--- a/arch/arm64/kernel/debug-monitors.c
+++ b/arch/arm64/kernel/debug-monitors.c
@@ -376,15 +376,13 @@ int aarch32_break_handler(struct pt_regs *regs)
 }
 NOKPROBE_SYMBOL(aarch32_break_handler);
 
-static int __init debug_traps_init(void)
+void __init debug_traps_init(void)
 {
 	hook_debug_fault_code(DBG_ESR_EVT_HWSS, single_step_handler, SIGTRAP,
 			      TRAP_TRACE, "single-step handler");
 	hook_debug_fault_code(DBG_ESR_EVT_BRK, brk_handler, SIGTRAP,
 			      TRAP_BRKPT, "ptrace BRK handler");
-	return 0;
 }
-arch_initcall(debug_traps_init);
 
 /* Re-enable single step for syscall restarting. */
 void user_rewind_single_step(struct task_struct *task)
diff --git a/arch/arm64/kernel/traps.c b/arch/arm64/kernel/traps.c
index cf402be5c573..8408e8670f2e 100644
--- a/arch/arm64/kernel/traps.c
+++ b/arch/arm64/kernel/traps.c
@@ -1047,11 +1047,11 @@ int __init early_brk64(unsigned long addr, unsigned int esr,
 	return bug_handler(regs, esr) != DBG_HOOK_HANDLED;
 }
 
-/* This registration must happen early, before debug_traps_init(). */
 void __init trap_init(void)
 {
 	register_kernel_break_hook(&bug_break_hook);
 #ifdef CONFIG_KASAN_SW_TAGS
 	register_kernel_break_hook(&kasan_break_hook);
 #endif
+	debug_traps_init();
 }
-- 
2.26.2.645.ge9eca65c58-goog

Re: [PATCH] arm64: Call debug_traps_init() from trap_init() to help early kgdb

[Will Deacon]

On Wed, May 13, 2020 at 04:06:37PM -0700, Douglas Anderson wrote:
> A new kgdb feature will soon land (kgdb_earlycon) that lets us run
> kgdb much earlier.  In order for everything to work properly it's
> important that the break hook is setup by the time we process
> "kgdbwait".
> 
> Right now the break hook is setup in debug_traps_init() and that's
> called from arch_initcall().  That's a bit too late since
> kgdb_earlycon really needs things to be setup by the time the system
> calls dbg_late_init().
> 
> We could fix this by adding call_break_hook() into early_brk64() and
> that works fine.  However, it's a little ugly.  Instead, let's just
> add a call to debug_traps_init() straight from trap_init().  There's
> already a documented dependency between trap_init() and
> debug_traps_init() and this makes the dependency more obvious rather
> than just relying on a comment.
> 
> NOTE: this solution isn't early enough to let us select the
> "ARCH_HAS_EARLY_DEBUG" KConfig option that is introduced by the
> kgdb_earlycon patch series.  That would only be set if we could do
> breakpoints when early params are parsed.  This patch only enables
> "late early" breakpoints, AKA breakpoints when dbg_late_init() is
> called.  It's expected that this should be fine for most people.
> 
> It should also be noted that if you crash you can still end up in kgdb
> earlier than debug_traps_init().  Since you don't need breakpoints to
> debug a crash that's fine.
> 
> Suggested-by: Will Deacon <will@kernel.org>
> Signed-off-by: Douglas Anderson <dianders@chromium.org>
> Cc: Catalin Marinas <catalin.marinas@arm.com>
> Cc: Will Deacon <will@kernel.org>
> ---
> This replaces the patch ("arm64: Add call_break_hook() to
> early_brk64() for early kgdb") in my recent kgdb series [1].  If I end
> up re-posting that series again I'll include this patch as a
> replacement, but I'm sending it separately to avoid spamming a pile of
> people another time with a 12-patch series.
> 
> Note that, because it doesn't select the "ARCH_HAS_EARLY_DEBUG"
> KConfig option it could be landed standalone.  However, it's still
> probably better to land together with that patch series.
> 
> If the kgdb_earlycon patch series lands without this patch then
> kgdbwait + kgdb_earlycon won't work well on arm64, but there would be
> no other bad side effects.
> 
> If this patch lands without the kgdb_earlycon patch series then there
> will be no known problems.
> 
> [1] https://lore.kernel.org/r/20200507130644.v4.5.I22067ad43e77ddfd4b64c2d49030628480f9e8d9@changeid
> 
>  arch/arm64/include/asm/debug-monitors.h | 2 ++
>  arch/arm64/kernel/debug-monitors.c      | 4 +---
>  arch/arm64/kernel/traps.c               | 2 +-
>  3 files changed, 4 insertions(+), 4 deletions(-)

[...]

Acked-by: Will Deacon <will@kernel.org>

I would prefer to take this via arm64, if possible, since we have quite lot
going in for 5.8, although I don't think this conflicts at the moment.

Daniel -- what do you want to do?

Will

Re: [PATCH] arm64: Call debug_traps_init() from trap_init() to help early kgdb

[Daniel Thompson]

On Fri, May 15, 2020 at 05:23:17PM +0100, Will Deacon wrote:
> On Wed, May 13, 2020 at 04:06:37PM -0700, Douglas Anderson wrote:
> > A new kgdb feature will soon land (kgdb_earlycon) that lets us run
> > kgdb much earlier.  In order for everything to work properly it's
> > important that the break hook is setup by the time we process
> > "kgdbwait".
> > 
> > Right now the break hook is setup in debug_traps_init() and that's
> > called from arch_initcall().  That's a bit too late since
> > kgdb_earlycon really needs things to be setup by the time the system
> > calls dbg_late_init().
> > 
> > We could fix this by adding call_break_hook() into early_brk64() and
> > that works fine.  However, it's a little ugly.  Instead, let's just
> > add a call to debug_traps_init() straight from trap_init().  There's
> > already a documented dependency between trap_init() and
> > debug_traps_init() and this makes the dependency more obvious rather
> > than just relying on a comment.
> > 
> > NOTE: this solution isn't early enough to let us select the
> > "ARCH_HAS_EARLY_DEBUG" KConfig option that is introduced by the
> > kgdb_earlycon patch series.  That would only be set if we could do
> > breakpoints when early params are parsed.  This patch only enables
> > "late early" breakpoints, AKA breakpoints when dbg_late_init() is
> > called.  It's expected that this should be fine for most people.
> > 
> > It should also be noted that if you crash you can still end up in kgdb
> > earlier than debug_traps_init().  Since you don't need breakpoints to
> > debug a crash that's fine.
> > 
> > Suggested-by: Will Deacon <will@kernel.org>
> > Signed-off-by: Douglas Anderson <dianders@chromium.org>
> > Cc: Catalin Marinas <catalin.marinas@arm.com>
> > Cc: Will Deacon <will@kernel.org>
> > ---
> > This replaces the patch ("arm64: Add call_break_hook() to
> > early_brk64() for early kgdb") in my recent kgdb series [1].  If I end
> > up re-posting that series again I'll include this patch as a
> > replacement, but I'm sending it separately to avoid spamming a pile of
> > people another time with a 12-patch series.
> > 
> > Note that, because it doesn't select the "ARCH_HAS_EARLY_DEBUG"
> > KConfig option it could be landed standalone.  However, it's still
> > probably better to land together with that patch series.
> > 
> > If the kgdb_earlycon patch series lands without this patch then
> > kgdbwait + kgdb_earlycon won't work well on arm64, but there would be
> > no other bad side effects.
> > 
> > If this patch lands without the kgdb_earlycon patch series then there
> > will be no known problems.
> > 
> > [1] https://lore.kernel.org/r/20200507130644.v4.5.I22067ad43e77ddfd4b64c2d49030628480f9e8d9@changeid
> > 
> >  arch/arm64/include/asm/debug-monitors.h | 2 ++
> >  arch/arm64/kernel/debug-monitors.c      | 4 +---
> >  arch/arm64/kernel/traps.c               | 2 +-
> >  3 files changed, 4 insertions(+), 4 deletions(-)
> 
> [...]
> 
> Acked-by: Will Deacon <will@kernel.org>
> 
> I would prefer to take this via arm64, if possible, since we have quite lot
> going in for 5.8, although I don't think this conflicts at the moment.
> 
> Daniel -- what do you want to do?

I'm very happy for you to take it!

On my side I hope to get the rest of the patchset into linux-next early
next week.


Daniel.

Re: [PATCH] arm64: Call debug_traps_init() from trap_init() to help early kgdb

[Will Deacon]

On Wed, 13 May 2020 16:06:37 -0700, Douglas Anderson wrote:
> A new kgdb feature will soon land (kgdb_earlycon) that lets us run
> kgdb much earlier.  In order for everything to work properly it's
> important that the break hook is setup by the time we process
> "kgdbwait".
> 
> Right now the break hook is setup in debug_traps_init() and that's
> called from arch_initcall().  That's a bit too late since
> kgdb_earlycon really needs things to be setup by the time the system
> calls dbg_late_init().
> 
> [...]

Applied to arm64 (for-next/misc), thanks!

[1/1] arm64: Call debug_traps_init() from trap_init() to help early kgdb
      https://git.kernel.org/arm64/c/b322c65f8ca3

Cheers,
-- 
Will

https://fixes.arm64.dev
https://next.arm64.dev
https://will.arm64.dev