I’ve connected two components using the reporter mechanism and attach_rom_dataspace to transmit signals along with a small amount of associated data. After running for some time, the system reports the following error: request resources: cap_quota=4.
Is there a more efficient or recommended way within Genode to transmit lightweight data alongside a signal?
and
Should capabilities be explicitly released after use in this context, or is their accumulation an expected behavior governed by the framework’s internal resource management policy?
Typically you’d instantiate the attached_rom, then use that during the app’s lifetime, then free it when the app tears down, freeing the used capability.
If the app is leaking caps, then
1.you could try to self-diagnose by sprinkling your code with “Genode::log( env.used_caps()… )” statements, to determine who/where is increasing the used capbabilities count, or
2. post the code here so that we can review it
I don’t remember seeing Genode leak caps by itself in my experiments (except a few years ago when calling pthread_create() but the culprit might have been me), so it should be relatively easy to find out what’s going on, if you create a minimum reproducible case
I’m using Genode’s native objects, specifically Genode::Reporter and Genode::Attach_rom_dataspace, so I would expect them to behave correctly. However, your point seems valid — I do use pthread in part of the code and interact with it. It appears that the target thread receives the signal but doesn’t release it afterward. Did I understand that correctly?
I was testing Genode’s notification mechanism within Qt5-based applications. My initial challenge was that notifications weren’t being received, which I first resolved using the pthread approach. Later, I learned that it’s also possible to connect a Qt5 signal to a notification, so I explored both methods to better understand the underlying processes.
I’m using libc to run Qt5, but I don’t perform any manual memory allocation—Qt5 handles all memory management internally.
When using the pthread method, I encountered a capability leak issue. I’m now trying to identify the root cause of this leak. At first, I suspected it might be related to the Genode::Attach_rom_dataspace object. However, as soon as I came across the term pthread in @ttcoder’s previous replies, I realized the memory leak likely stems from my own implementation. I’m currently investigating the exact cause of the leak.
