inline functions funpack(), fpack() considered to large for inlining

Hi Michael,

I understand your point but I've not been able to reproduce it. Do you have a smallish example source file you post that demonstrates the remark.

Thanks,

Stuart.

Hi Stuart,

find attached an example project. Please compile target "DebugNWC".

Regards,

Michael

Hi,

using the example you provided, I could not reproduce the remarks you are seeing, either with Update 8 or with the latest update, Update 9.1. The Compiler issues the following two remarks when compiling this example in Debug NWC:

".\inline-funpack.c", line 20 (col. 29): cc1461: {D} remark: inlining funpack

".\inline-funpack.c", line 22 (col. 27): cc1461: {D} remark: inlining fpack

We may need to check your version information more closely. As the VisualDSP++ "Product Info" HTML file contains licensing information, could I ask you to contact private support through the link below, and include a link to this thread:

http://www.analog.com/processor/support

Regards,

Craig.

Hi Michael,

The help text for cc1461 is wrong. The cc1461 remark is issued as an informational message to indicate that the compiler has inlined a function. It's not conditional on any costing or gain considerations and as far as we can determine it never has been (which is a bit of a mystery we need to investigate).

For example, the same remark is issued for the simplest of inline functions as seen below:

[edin-forres:~] cat t.c
inline void foo(void) {
}

void bar(void) {
foo();
}

[edin-forres:~] ccblkfn.exe -Wremarks -O t.c -S
"t.c", line 5 (col. 6): cc1461: {D} remark: inlining foo

We'll fix the cc1461 text in a future update of the compiler message documentation (our reference is TAR-47014) but rest assured we are certain that you can safely ignore it.

Regards,

Stuart.

Dear Craig,

the remark you got is the behaviour, I described in my first posting.

You mentioned, you got the following remarks:

".\inline-funpack.c", line 20 (col. 29): cc1461: {D} remark: inlining funpack

".\inline-funpack.c", line 22 (col. 27): cc1461: {D} remark: inlining fpack

VisualDSP help tells me for cc1461:

Compiler Remark: function call inlined despite compiler considering unsuitable for inlining

Description: This remark is raised when the compiler has been forced to inline a function call when the compiler considers the function unsuitably large for inlining. This occurs when the function in question is defined using #pragma always_inline or the –always-inline switch is passed to the compiler.

So here is my question again: Why does the C-compiler consider funpack()/fpack() unsuitable for inlining? They are intrinsic functions with (in theory) only one assembly instruction.

Best regards,

Michael

This question has been assumed as answered either offline via email or with a multi-part answer. This question has now been closed out. If you have an inquiry related to this topic please post a new question in the applicable product forum.

Thank you,
EZ Admin