| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <20240207035457.GA19804@DESKTOP-KA7F9LU.localdomain>
Date: Wed, 7 Feb 2024 09:24:57 +0530
From: Vimal Kumar <vimal.kumar32@...il.com>
To: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
Cc: "Rafael J. Wysocki" <rafael@...nel.org>, Pavel Machek <pavel@....cz>,
Len Brown <len.brown@...el.com>, linux-pm@...r.kernel.org,
linux-kernel@...r.kernel.org, chinmoyghosh2001@...il.com,
badolevishal1116@...il.com, mintupatel89@...il.com
Subject: Re: [PATCH v4] PM / sleep: Mechanism to find source aborting kernel
suspend transition
On Mon, Feb 05, 2024 at 07:33:17PM +0000, Greg Kroah-Hartman wrote:
> On Mon, Feb 05, 2024 at 10:37:45PM +0530, Vimal Kumar wrote:
> > Sometimes kernel suspend transitions can be aborted unconditionally by
> > manipulating pm_abort_suspend value using "hard" wakeup triggers or
> > through "pm_system_wakeup()".
> >
> > There is no way to trace the source path of module or subsystem which
> > aborted the suspend transitions. This change will create a list of
> > wakeup sources aborting suspend in progress through "hard" events as
> > well as subsytems aborting suspend using "pm_system_wakeup()".
> >
> > Example: Existing suspend failure logs:
> > [ 349.708359] PM: Some devices failed to suspend, or early wake event detected
> > [ 350.327842] PM: suspend exit
> >
> > Suspend failure logs with this change:
> > [ 518.761835] PM: Some devices failed to suspend, or early wake event detected
> > [ 519.486939] PM: wakeup source or subsystem uart_suspend_port aborted suspend
> > [ 519.500594] PM: suspend exit
> >
> > Here we can clearly identify the module triggerring abort suspend.
> >
> > Co-developed-by: Chinmoy Ghosh <chinmoyghosh2001@...il.com>
> > Signed-off-by: Chinmoy Ghosh <chinmoyghosh2001@...il.com>
> > Co-developed-by: Mintu Patel <mintupatel89@...il.com>
> > Signed-off-by: Mintu Patel <mintupatel89@...il.com>
> > Co-developed-by: Vishal Badole <badolevishal1116@...il.com>
> > Signed-off-by: Vishal Badole <badolevishal1116@...il.com>
> > Signed-off-by: Vimal Kumar <vimal.kumar32@...il.com>
> > ---
> > Changes in v4:
> > - Changed GFP_KERNEL flag to GFP_ATOMIC
> > - Changed mutex_lock to raw_spin_lock
>
> why raw?
>
As mutex_lock might sleep, we need to use lock that is suitable for usages in atomic context. raw_spin_lock is already being used for other list in
this driver, so I used the same. If suggested we can switch to spin_lock_irqsave as well.
>
> > ---
> > drivers/base/power/wakeup.c | 100 +++++++++++++++++++++++++++++++++++-
> > 1 file changed, 99 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/base/power/wakeup.c b/drivers/base/power/wakeup.c
> > index a917219feea6..b04794557eef 100644
> > --- a/drivers/base/power/wakeup.c
> > +++ b/drivers/base/power/wakeup.c
> > @@ -73,6 +73,16 @@ static struct wakeup_source deleted_ws = {
> >
> > static DEFINE_IDA(wakeup_ida);
> >
> > +#ifdef CONFIG_PM_DEBUG
> > +static DEFINE_RAW_SPINLOCK(pm_abort_suspend_list_lock);
> > +
> > +struct pm_abort_suspend_source {
> > + struct list_head list;
> > + char *source_triggering_abort_suspend;
> > +};
> > +static LIST_HEAD(pm_abort_suspend_list);
> > +#endif
> > +
> > /**
> > * wakeup_source_create - Create a struct wakeup_source object.
> > * @name: Name of the new wakeup source.
> > @@ -575,6 +585,54 @@ static void wakeup_source_activate(struct wakeup_source *ws)
> > trace_wakeup_source_activate(ws->name, cec);
> > }
> >
> > +#ifdef CONFIG_PM_DEBUG
>
> Please do not add #ifdef to .c files, this makes this file even messier.
>
> > @@ -590,8 +648,13 @@ static void wakeup_source_report_event(struct wakeup_source *ws, bool hard)
> > if (!ws->active)
> > wakeup_source_activate(ws);
> >
> > - if (hard)
> > + if (hard) {
> > +#ifdef CONFIG_PM_DEBUG
> > + if (pm_suspend_target_state != PM_SUSPEND_ON)
> > + pm_abort_suspend_source_add(ws->name);
> > +#endif
>
> Especially for stuff like this, if you write your .h files properly, no
> #ifdef are needed.
>
Thanks. We can move it to .h file.
>
>
>
> > pm_system_wakeup();
> > + }
> > }
> >
> > /**
> > @@ -893,12 +956,47 @@ bool pm_wakeup_pending(void)
> > pm_print_active_wakeup_sources();
> > }
> >
> > +#ifdef CONFIG_PM_DEBUG
> > + if (atomic_read(&pm_abort_suspend) > 0) {
> > + struct pm_abort_suspend_source *info;
> > +
> > + raw_spin_lock_irqsave(&pm_abort_suspend_list_lock, flags);
> > + list_for_each_entry(info, &pm_abort_suspend_list, list) {
> > + pm_pr_dbg("wakeup source or subsystem %s aborted suspend\n",
> > + info->source_triggering_abort_suspend);
> > + }
> > + raw_spin_unlock_irqrestore(&pm_abort_suspend_list_lock, flags);
> > + pm_abort_suspend_list_clear();
> > + }
> > +#endif
> > +
> > return ret || atomic_read(&pm_abort_suspend) > 0;
> > }
> > EXPORT_SYMBOL_GPL(pm_wakeup_pending);
> >
> > void pm_system_wakeup(void)
> > {
> > +
> > +#ifdef CONFIG_PM_DEBUG
> > +#ifdef CONFIG_DEBUG_INFO
> > + if (pm_suspend_target_state != PM_SUSPEND_ON) {
> > + char *source_name = kasprintf(GFP_ATOMIC,
> > + "%ps",
> > + __builtin_return_address(0));
> > + if (!source_name)
> > + goto exit;
> > +
> > + if (strcmp(source_name, "pm_wakeup_ws_event"))
> > + pm_abort_suspend_source_add(source_name);
> > +
> > + kfree(source_name);
> > + }
> > +exit:
> > +#else
> > + if (pm_suspend_target_state != PM_SUSPEND_ON)
> > + pm_pr_dbg("Some wakeup source or subsystem aborted suspend\n");
> > +#endif
> > +#endif
>
> Would you want to maintain this #ifdef nesting mess for the next 20
> years? Please do not do this.
>
I was hoping if we can remove the "CONFIG_PM_DEBUG" as this functionality can exist by default as well.
We can avoid nesting and usage for CONFIG_DEBUG_INFO move to .h files.
Please share your perspective on this.
> thanks,
>
> greg k-h
Warm Regards,
Vimal Kumar
Powered by blists - more mailing lists