Re: [PATCH v2 16/17] driver core: Refactor fw_devlink feature

From: Marc Zyngier
Date: Fri Dec 11 2020 - 13:05:13 EST


On 2020-12-11 17:51, Saravana Kannan wrote:
On Fri, Dec 11, 2020 at 8:34 AM Robin Murphy <robin.murphy@xxxxxxx> wrote:

On 2020-12-11 14:11, Qian Cai wrote:
> On Fri, 2020-11-20 at 18:02 -0800, Saravana Kannan wrote:
>> The current implementation of fw_devlink is very inefficient because it
>> tries to get away without creating fwnode links in the name of saving
>> memory usage. Past attempts to optimize runtime at the cost of memory
>> usage were blocked with request for data showing that the optimization
>> made significant improvement for real world scenarios.
>>
>> We have those scenarios now. There have been several reports of boot
>> time increase in the order of seconds in this thread [1]. Several OEMs
>> and SoC manufacturers have also privately reported significant
>> (350-400ms) increase in boot time due to all the parsing done by
>> fw_devlink.
>>
>> So this patch uses all the setup done by the previous patches in this
>> series to refactor fw_devlink to be more efficient. Most of the code has
>> been moved out of firmware specific (DT mostly) code into driver core.
>>
>> This brings the following benefits:
>> - Instead of parsing the device tree multiple times during bootup,
>> fw_devlink parses each fwnode node/property only once and creates
>> fwnode links. The rest of the fw_devlink code then just looks at these
>> fwnode links to do rest of the work.
>>
>> - Makes it much easier to debug probe issue due to fw_devlink in the
>> future. fw_devlink=on blocks the probing of devices if they depend on
>> a device that hasn't been added yet. With this refactor, it'll be very
>> easy to tell what that device is because we now have a reference to
>> the fwnode of the device.
>>
>> - Much easier to add fw_devlink support to ACPI and other firmware
>> types. A refactor to move the common bits from DT specific code to
>> driver core was in my TODO list as a prerequisite to adding ACPI
>> support to fw_devlink. This series gets that done.
>>
>> [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@xxxxxx/
>> Signed-off-by: Saravana Kannan <saravanak@xxxxxxxxxx>
>> Tested-by: Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx>
>> Tested-by: Grygorii Strashko <grygorii.strashko@xxxxxx>
>
> Reverting this commit and its dependency:
>
> 2d09e6eb4a6f driver core: Delete pointless parameter in fwnode_operations.add_links
>
> from today's linux-next fixed a boot crash on an arm64 Thunder X2 server.

Since the call stack implicates the platform-device-wrangling we do in
IORT code I took a quick look; AFAICS my guess would be it's blowing up
trying to walk a zeroed list head since "driver core: Add fwnode_init()"
missed acpi_alloc_fwnode_static().

Thanks Robin. I'm pretty sure this is the reason. I thought I fixed
all ACPI cases, but clearly I missed this one. I'll send out a patch
for this today. If you think there are any other places I missed
please let me know. I'll try some git grep foo to see if I missed any
other instances of fwnode ops being set.

Yup, that fixed it here (QDF2400).

Thanks,

M.

diff --git a/include/linux/acpi.h b/include/linux/acpi.h
index 39263c6b52e1..2630c2e953f7 100644
--- a/include/linux/acpi.h
+++ b/include/linux/acpi.h
@@ -55,7 +55,7 @@ static inline struct fwnode_handle *acpi_alloc_fwnode_static(void)
if (!fwnode)
return NULL;

- fwnode->ops = &acpi_static_fwnode_ops;
+ fwnode_init(fwnode, &acpi_static_fwnode_ops);

return fwnode;
}



--
Jazz is not dead. It just smells funny...