-
Notifications
You must be signed in to change notification settings - Fork 203
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Clarify/clean up spacecraft/cpu names/ids #710
Comments
The CFE_CPU_NAME, CFE_CPU_ID, and CFE_SPACECRAFT_ID should already be deprecated. The values from CMake are the ones which actually drive the return value of The recommended practice should be for apps to call these PSP routines to get the values at runtime, not to rely on the value in the header file. Now that CMake is being used for all, the CMake definitions should be there, so we should deprecate all header file constants. Unfortunately, there is no |
Correct. I'm more surprised about use of the PLATFORM/MISSION values in table logic. Didn't dig into it though. |
Yeah, we might have a bug here in that tables will get the wrong value if they use the header file define. Probably need to remove from these headers and add as "-D" options when invoking the compiler as part of table file builds. |
Quick check shows these TBL Values are only used depending on CFE_PLATFORM_TBL_VALID_SCID_COUNT and CFE_PLATFORM_TBL_VALID_PRID_COUNT being greater than 0. Default config is zero. Also a case of the code getting "compiled out" when its zero, so it could be broken. |
Also note about table file builds that the actual value in the header (IIRC) comes from elf2cfetbl command line parameters, which comes from CMake, so its OK. I don't know of any tables that actually contain the CPU ID / Spacecraft ID in the table content. |
I've noticed it would probably be desirable to be able to override the CPUID manually, rather than having them sequentially assigned by TGTID, in scenarios where multiple integration repositories are used to build the spacecraft software and they all need to communicate over SBN(g). |
Ultimately the PSP determines the CPU ID at runtime. The value from CMake is just a default. For instance the pc-linux PSP allows it to be overridden on the command line via an argument. Other FSW target PSP's could do something similar. Table builds are a different story- if you also need to put the CPU ID in a table then this would be a new feature/enhancement. |
Ping @excaliburtb |
So, no one should not rely on the contents of cfe_platform_cfg.h directly from an app level but from an integration level, the default values compiled into the PSP should probably come from the header, not from cmake. In other words, I would want CFE_PSP_GetProcessorId() to return CFE_PLATFORM_CPU_ID by default. The advantage of the header is that it clearly defines those values and is not obfuscated by cmake variable or build scheme. If you want to provide an optional cmake variable that allows the user to override the header content, I guess you could do that. |
Note that there is (currently) no requirement for a unique/different Also note that Conversly - If we fixed #265 to allow independently configuring the number rather than assuming consecutive numbering, wouldn't that also address this issue? In that case, the CPU numeric IDs would be specified in targets.cmake right where the name is specified, which seems quite logical to me. They just wouldn't have to match the CMake IDs anymore. |
Yes, fixing #265 would help a lot. The point is, don't decide for me what the default cpuid is with no way to set it myself. |
Okay, there's one more way to look at this that shouldn't require me waiting for #265. Right now the default CPU ID is actually built into target_config.c which is not in the PSP itself. It's in CFE. So, I need to be able to specify, either by platform_cfg.h or a cmake variable what is used to set that default. I'm assuming target_config.c is not considered a "shared" library between targets. |
Add support for a TGTx_PROCESSOR_ID directive, which allows one to set the default value returned by CFE_PSP_GetProcessorId() function, rather than assuming the index value from CMake.
Add support for a TGTx_PROCESSOR_ID directive, which allows one to set the default value returned by CFE_PSP_GetProcessorId() function, rather than assuming the index value from CMake.
Add support for a TGTx_PROCESSOR_ID directive, which allows one to set the default value returned by CFE_PSP_GetProcessorId() function, rather than assuming the index value from CMake.
Fix #710, Specify ProcessorID in targets.cmake
Add support for a TGTx_PROCESSOR_ID directive, which allows one to set the default value returned by CFE_PSP_GetProcessorId() function, rather than assuming the index value from CMake.
Is your feature request related to a problem? Please describe.
Defined in platform config:
Defined in mission config:
Defined by cmake:
Describe the solution you'd like
Remove/deprecate unused and/or clarify use/intent. Looks like spacecraft ID may actually be defined differently (42 vs 0x42).
Describe alternatives you've considered
Stay confused.
Additional context
Table use looks like it could lead to inconsistencies.
Requester Info
Jacob Hageman - NASA/GSFC, triggered by @johnphamngc comments on nasa/PSP#154
The text was updated successfully, but these errors were encountered: