Re: Failures on erc32 and psim

2017-10-08 Thread Sebastian Huber
One issue was related due to the new 64-bit time_t, e.g. (time_t) (uint32_t) -1 is greater than zero now. On 09/10/17 02:25, Joel Sherrill wrote: Hi psx05 and capture timeout  on psim and erc32. I think psx05 is hanging up and not a random timeout. capture appears to be miscategorized. It is

Re: RFC: confdefs.h macro for C11 threads

2017-10-08 Thread Sebastian Huber
On 08/10/17 22:31, Joel Sherrill wrote: Hi We have a precedence of presenting an abstracted notion of configuring an object. Users should not have to know how something is implemented in RTEMS. We don't address the configuration of C11 threading objects in this way. Given the implementation, I

Re: RFC: confdefs.h macro for C11 threads

2017-10-08 Thread Sebastian Huber
On 09/10/17 02:32, Chris Johns wrote: [...] How does a user use C11 threads on RTEMS for real applications? When I last looked I could not see a way to configure any runtime parameters like stack or priority. You can do this in an implementation-specifc way via the native_handle(): http://en.

Re: RFC: confdefs.h macro for C11 threads

2017-10-08 Thread Chris Johns
On 09/10/2017 07:31, Joel Sherrill wrote: > Hi > > We have a precedence of presenting an abstracted notion of > configuring an object. Users should not have to know how something > is implemented in RTEMS. We don't address the configuration of > C11 threading objects in this way. > > Given the i

Failures on erc32 and psim

2017-10-08 Thread Joel Sherrill
Hi psx05 and capture timeout on psim and erc32. I think psx05 is hanging up and not a random timeout. capture appears to be miscategorized. It is interactive. psim has failures beyond those on erc32. Failures: psxsignal01.exe spfatal31.exe Timeouts: dl05.exe capture.exe unlimited.exe psx

RFC: confdefs.h macro for C11 threads

2017-10-08 Thread Joel Sherrill
Hi We have a precedence of presenting an abstracted notion of configuring an object. Users should not have to know how something is implemented in RTEMS. We don't address the configuration of C11 threading objects in this way. Given the implementation, I proposes we add CONFIGURE_C11_MAXIMUM_THRE

RFC: Remove confdefs.h "*HAS_OWN*" support

2017-10-08 Thread Joel Sherrill
Hi Since Sebastian changed the language in Configuring a System to make it mandatory to use confdefs.h, I think we should remove the *HAS_OWN* configuration parameters. We haven't recommended anyone bypass confdefs.h and define their own instances of configuration tables and variables in many ma