Update grut699-vxworks6 cfe_psp_memory.c per white box unit testing results

[skliper]

During white box testing a number of functions were identified which return CFE_PSP_ERROR rather than CFE_PSP_INVALID_POINTER for NULL pointers. Functions include:
CFE_PSP_GetCDSSize
CFE_PSP_WriteToCDS
CFE_PSP_ReadFromCDS
CFE_PSP_GetResetArea
CFE_PSP_GetUserReservedArea
CFE_PSP_GetVolatileDiskMem
CFE_PSP_GetKernelTextSegmentInfo
CFE_PSP_GetCFETextSegmentInfo

Other issues:
Several functions take two pointer arguments, but only check if one is a NULL pointer.

Should check calling functions to see if changes to the return values may potentially cause issues.

Issues identified during #14 white box testing commit: [changeset:612f00f3]

[skliper]

Imported from trac issue 33. Created by dasp on 2015-08-11T17:18:16, last modified: 2019-05-06T11:43:03

[skliper]

Trac comment by sduran on 2015-08-18 10:44:19:

Changing this from _ERROR to _INVALID_POINTER is effectively an API change. Hold off making changes for the return code until an analysis of all of the calling functions is made. It could potentially result in a large number of changes.

However, the null pointer checks, to prevent potential crashes would be good to add.

[skliper]

Trac comment by abrown4 on 2015-08-18 12:57:04:

commit: [changeset:532afb2] Trac #37 Added null pointer checks to cfe_psp_memory.c (no API change).
Updated header comments and unit tests for current return values.

[skliper]

Trac comment by abrown4 on 2015-08-18 13:03:53:

In CFE_PSP_GetCFETextSegmentInfo(), line 421, if the moduleInfoGet() fails, the status is checked... but both paths have CFE_PSP_GetCFETextSegmentInfo() return CFE_PSP_SUCCESS. This is a problem. This function is called from cfe_es_task.c, where it does check the return status vs. OS_SUCCESS. The SizeOfCFESegment is assumed to be good with OS_SUCCESS.

commit: [changeset:b0fe8a4] Trac #37 Fixed inproper return type for CFE_PSP_GetCFETextSegmentInfo().

[skliper]

Trac comment by abrown4 on 2015-08-18 13:11:13:

In function CFE_PSP_GetKernelTextSegmentInfo(), white box unit testing, commit [changeset:612f00f3] through [changeset:b0fe8a4], shows that there is no sanity-checking the values returned by GetWrsKernelTextStart() and GetWrsKernelTextEnd().

Do we want to add this functionality or not to CFE_PSP_GetKernelTextSegmentInfo()?

[skliper]

Trac comment by sduran on 2015-08-18 15:15:01:

There is nothing to sanity check, unless you already know what the addresses should be, but you can't until you call these functions...

/*
** Function: GetWrsKernelTextStart
** Purpose: This function returns the start address of the kernel code.
**
*/
unsigned int GetWrsKernelTextStart (void)
{
return (unsigned int) &wrs_kernel_text_start;
}

/*
** Function: GetWrsKernelTextEnd
** Purpose: This function returns the end address of the kernel code.
**
*/
unsigned int GetWrsKernelTextEnd (void)
{
return (unsigned int) &wrs_kernel_text_end;
}

[skliper]

Trac comment by abrown4 on 2015-08-18 15:21:38:

Alan's WB test logic showed an implicit assumption that GetWrsKernelTextEnd() is expected to return a value greater than what you get from GetWrsKernelTextStart().

[skliper]

Trac comment by abrown4 on 2015-08-18 15:22:02:

(merged to #14 branch)

[skliper]

Trac comment by jhageman on 2019-05-06 11:43:03:

Specific to non-community PSP so closing as invalid. Note changes were merged to techdev-pc-linux-uei, techdev-sp0-vxworks6.9 branches (but not techdev-ut699 for some reason). Not in framework release.