| Message ID | 20230419155341.v8.7.I21d92f8974c8e4001a5982fea6c98da1bed33ef5@changeid (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [v8,01/10] arm64: Add framework to turn IPI as NMI | expand |
Hi, On Fri, May 12, 2023 at 6:49 AM Daniel Thompson <daniel.thompson@linaro.org> wrote: > > On Wed, Apr 19, 2023 at 03:56:01PM -0700, Douglas Anderson wrote: > > From: Sumit Garg <sumit.garg@linaro.org> > > > > Add a new API kgdb_smp_call_nmi_hook() to expose default CPUs roundup > > mechanism to a particular archichecture as a runtime fallback if it > > detects to not support NMI roundup. > > > > Currently such an architecture example is arm64 supporting pseudo NMIs > > feature which is only available on platforms which have support for GICv3 > > or later version. > > > > Signed-off-by: Sumit Garg <sumit.garg@linaro.org> > > Tested-by: Chen-Yu Tsai <wens@csie.org> > > Signed-off-by: Douglas Anderson <dianders@chromium.org> > > --- > > > > (no changes since v1) > > > > include/linux/kgdb.h | 12 ++++++++++++ > > kernel/debug/debug_core.c | 8 +++++++- > > 2 files changed, 19 insertions(+), 1 deletion(-) > > > > diff --git a/include/linux/kgdb.h b/include/linux/kgdb.h > > index 258cdde8d356..87713bd390f3 100644 > > --- a/include/linux/kgdb.h > > +++ b/include/linux/kgdb.h > > @@ -199,6 +199,18 @@ kgdb_arch_handle_qxfer_pkt(char *remcom_in_buffer, > > > > extern void kgdb_call_nmi_hook(void *ignored); > > > > +/** > > + * kgdb_smp_call_nmi_hook - Provide default fallback mechanism to > > + * round-up CPUs > > + * > > + * If you're using the default implementation of kgdb_roundup_cpus() > > + * this function will be called. And if an arch detects at runtime to > > + * not support NMI based roundup then it can fallback to default > > + * mechanism using this API. > > + */ > > + > > +extern void kgdb_smp_call_nmi_hook(void); > > Concept looks sensible but this is a terrible name for aa command to > round up the CPUs using smp_call... functions. Whilst it is true it that > kgdb_roundup_cpus() does use kgdb_call_nmi_hook() internally that > doesn't mean we should name functions after it. They should be named > after what they are do, not how they do it. > > Something more like kgdb_roundup_cpus_with_smp_call() would be a much > better name. Sounds good. I'm happy to spin with this rename, though I was kinda hoping to drop ${SUBJECT} patch if folks were OK with patch #10 in this series [1]. I personally think that's the right way to go but it's unclear to me if arm64 maintainers will think it's a hack (despite the fact that arm32 implements the "nmi" functions with regular IPIs). For now, maybe I'll think positive thoughts and hope that folks will have the time to review the series soon. If another few weeks go by then I'll send a v9 with just your comments addressed. If nothing else, maybe you can land the kgdb parts in a tree targeting v6.5 and then when arm64 folks have the bandwidth then it will be easier to get them landed. [1] https://lore.kernel.org/r/20230419155341.v8.10.Ic3659997d6243139d0522fc3afcdfd88d7a5f030@changeid -Doug
Hi, On Mon, May 15, 2023 at 4:21 PM Doug Anderson <dianders@chromium.org> wrote: > > Hi, > > On Fri, May 12, 2023 at 6:49 AM Daniel Thompson > <daniel.thompson@linaro.org> wrote: > > > > On Wed, Apr 19, 2023 at 03:56:01PM -0700, Douglas Anderson wrote: > > > From: Sumit Garg <sumit.garg@linaro.org> > > > > > > Add a new API kgdb_smp_call_nmi_hook() to expose default CPUs roundup > > > mechanism to a particular archichecture as a runtime fallback if it > > > detects to not support NMI roundup. > > > > > > Currently such an architecture example is arm64 supporting pseudo NMIs > > > feature which is only available on platforms which have support for GICv3 > > > or later version. > > > > > > Signed-off-by: Sumit Garg <sumit.garg@linaro.org> > > > Tested-by: Chen-Yu Tsai <wens@csie.org> > > > Signed-off-by: Douglas Anderson <dianders@chromium.org> > > > --- > > > > > > (no changes since v1) > > > > > > include/linux/kgdb.h | 12 ++++++++++++ > > > kernel/debug/debug_core.c | 8 +++++++- > > > 2 files changed, 19 insertions(+), 1 deletion(-) > > > > > > diff --git a/include/linux/kgdb.h b/include/linux/kgdb.h > > > index 258cdde8d356..87713bd390f3 100644 > > > --- a/include/linux/kgdb.h > > > +++ b/include/linux/kgdb.h > > > @@ -199,6 +199,18 @@ kgdb_arch_handle_qxfer_pkt(char *remcom_in_buffer, > > > > > > extern void kgdb_call_nmi_hook(void *ignored); > > > > > > +/** > > > + * kgdb_smp_call_nmi_hook - Provide default fallback mechanism to > > > + * round-up CPUs > > > + * > > > + * If you're using the default implementation of kgdb_roundup_cpus() > > > + * this function will be called. And if an arch detects at runtime to > > > + * not support NMI based roundup then it can fallback to default > > > + * mechanism using this API. > > > + */ > > > + > > > +extern void kgdb_smp_call_nmi_hook(void); > > > > Concept looks sensible but this is a terrible name for aa command to > > round up the CPUs using smp_call... functions. Whilst it is true it that > > kgdb_roundup_cpus() does use kgdb_call_nmi_hook() internally that > > doesn't mean we should name functions after it. They should be named > > after what they are do, not how they do it. > > > > Something more like kgdb_roundup_cpus_with_smp_call() would be a much > > better name. > > Sounds good. I'm happy to spin with this rename, though I was kinda > hoping to drop ${SUBJECT} patch if folks were OK with patch #10 in > this series [1]. I personally think that's the right way to go but > it's unclear to me if arm64 maintainers will think it's a hack > (despite the fact that arm32 implements the "nmi" functions with > regular IPIs). > > For now, maybe I'll think positive thoughts and hope that folks will > have the time to review the series soon. If another few weeks go by > then I'll send a v9 with just your comments addressed. If nothing > else, maybe you can land the kgdb parts in a tree targeting v6.5 and > then when arm64 folks have the bandwidth then it will be easier to get > them landed. > > [1] https://lore.kernel.org/r/20230419155341.v8.10.Ic3659997d6243139d0522fc3afcdfd88d7a5f030@changeid Breadcrumbs: I've dropped this patch and several others in v9 [1] by embracing the idea of using a normal IPI as a fallback. [1] https://lore.kernel.org/r/20230601213440.2488667-1-dianders@chromium.org/
diff --git a/include/linux/kgdb.h b/include/linux/kgdb.h index 258cdde8d356..87713bd390f3 100644 --- a/include/linux/kgdb.h +++ b/include/linux/kgdb.h @@ -199,6 +199,18 @@ kgdb_arch_handle_qxfer_pkt(char *remcom_in_buffer, extern void kgdb_call_nmi_hook(void *ignored); +/** + * kgdb_smp_call_nmi_hook - Provide default fallback mechanism to + * round-up CPUs + * + * If you're using the default implementation of kgdb_roundup_cpus() + * this function will be called. And if an arch detects at runtime to + * not support NMI based roundup then it can fallback to default + * mechanism using this API. + */ + +extern void kgdb_smp_call_nmi_hook(void); + /** * kgdb_roundup_cpus - Get other CPUs into a holding pattern * diff --git a/kernel/debug/debug_core.c b/kernel/debug/debug_core.c index d5e9ccde3ab8..14d40a7d6a4b 100644 --- a/kernel/debug/debug_core.c +++ b/kernel/debug/debug_core.c @@ -238,7 +238,7 @@ NOKPROBE_SYMBOL(kgdb_call_nmi_hook); static DEFINE_PER_CPU(call_single_data_t, kgdb_roundup_csd) = CSD_INIT(kgdb_call_nmi_hook, NULL); -void __weak kgdb_roundup_cpus(void) +void kgdb_smp_call_nmi_hook(void) { call_single_data_t *csd; int this_cpu = raw_smp_processor_id(); @@ -269,6 +269,12 @@ void __weak kgdb_roundup_cpus(void) kgdb_info[cpu].rounding_up = false; } } +NOKPROBE_SYMBOL(kgdb_smp_call_nmi_hook); + +void __weak kgdb_roundup_cpus(void) +{ + kgdb_smp_call_nmi_hook(); +} NOKPROBE_SYMBOL(kgdb_roundup_cpus); #endif