invalid signal-context capability

[Alexander Boettcher]

Hello,

On 7/18/22 10:51 AM, Stefan Thöni wrote:
> I'm debugging an "invalid signal-context capability" but I can't 
> pinpoint the bug.
> 
>  From my understanding an "invalid signal-context capability" happens 
> when the signal handler is destroyed before the signal is processed. I 
> have therefore carefully looked at the lifetime of all signal handlers 
> and their corresponding signals, refactoring code to allow signals to be 
> processed before destroying signal handlers.
> 
> Is there another condition that triggers an "invalid signal-context 
> capability" besides destroying the signal handler before any 
> corresponding signal is processed?

Is in your case the cap in _deliver_from_ep already invalid (add cap to 
the error message or check cap.valid()/cap.local_name()) or looks the 
cap valid and the context is not found ?

> Does anyone have any tips on how to find which signal handler caused an 
> "invalid signal-context capability" or where the signal originated?

When the error case triggers, you may extend the local Rpc_deliver with 
a error return value, which you may check on the caller side 
(submit(cap,cnt)). If the error code is set you may print the thread 
name of the caller and make a Genode::backtrace() to find the source.

Hope it helps,

Alex.

-- 
Alexander Boettcher
Genode Labs

https://www.genodians.org - https://www.genode.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: OpenPGP_0xBB4DE7FF9FF6ED5E.asc
Type: application/pgp-keys
Size: 3163 bytes
Desc: OpenPGP public key
URL: <http://lists.genode.org/pipermail/users/attachments/20220719/16e89fc6/attachment.key>