SPRESENSE closed issue #11236: Fail to nxtask_delete by update of cancel point
logic
URL: https://github.com/apache/nuttx/issues/11236
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific
SPRESENSE commented on issue #11236:
URL: https://github.com/apache/nuttx/issues/11236#issuecomment-1823913771
Close as completed.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific
SPRESENSE commented on issue #11236:
URL: https://github.com/apache/nuttx/issues/11236#issuecomment-1821988026
@pussuw
I could confirm that this issue was fixed by
https://github.com/apache/nuttx/pull/11241.
--
This is an automated message from the Apache Git Service.
To respond to
pussuw commented on issue #11236:
URL: https://github.com/apache/nuttx/issues/11236#issuecomment-1820890026
@SPRESENSE
This should fix your issue https://github.com/apache/nuttx/pull/11241/
Sorry about the regression, did not test it well enough..
--
This is an automated
pussuw commented on issue #11236:
URL: https://github.com/apache/nuttx/issues/11236#issuecomment-1820774823
Once the regression is patched I must think if this "peeking to another
process's stack" is a security concern which requires re-designing how
nxnotify_cancellation is handled.
pussuw commented on issue #11236:
URL: https://github.com/apache/nuttx/issues/11236#issuecomment-1820770939
Hi @SPRESENSE funnily enough, I found this exact line myself just 5 seconds
ago :)
I will think of a patch to fix this. The security test (groups are equal)
does not make
SPRESENSE commented on issue #11236:
URL: https://github.com/apache/nuttx/issues/11236#issuecomment-1820766516
@pussuw Thanks for your investigation.
My conditions are:
* CONFIG_CANCELLATION_POINTS=n
* CONFIG_SCHED_THREAD_LOCAL=n
I found that `nxsched_get_stackinfo()`
pussuw commented on issue #11236:
URL: https://github.com/apache/nuttx/issues/11236#issuecomment-1820670872
One last thing that comes to mind, my assumption was that the stack frame
created for TLS is memset to 0 always
```
int tls_init_info(FAR struct tcb_s *tcb)
{
FAR
pussuw commented on issue #11236:
URL: https://github.com/apache/nuttx/issues/11236#issuecomment-1820654033
@SPRESENSE sorry about the regression..
Yes, I'll start looking immediately. I can not find an example config with
the sensor SENSORS_CXD5610_GNSS enabled, so a few questions:
xiaoxiang781216 commented on issue #11236:
URL: https://github.com/apache/nuttx/issues/11236#issuecomment-1820588238
@pussuw could you take a look?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go
SPRESENSE opened a new issue, #11236:
URL: https://github.com/apache/nuttx/issues/11236
This pull request https://github.com/apache/nuttx/pull/11165 causes the task
to fail to delete.
There is a gnss sensor driver, which creates a task with `open()` and
deletes it with `close()`.
11 matches
Mail list logo
