Some cleanup in secure session manager code

[pan-apple]

Problem

SecureSessionMgr::SendMessage() function has grown too big in size, and needs some refactoring.

Summary of Changes

• Moved code to encrypt message to its own function
• Moved code to extract packet header from an pre-encrypted message out of the common function
• Reorganized code to reduce conditional statements and blocks
• Fixed a memory leak (if mTransportMgr->SendMessage failed, it could have leaked the retained buffer)

[github-actions]

Size increase report for "esp32-example-build" from cb23e0c

File	Section	File	VM
chip-all-clusters-app.elf	.flash.text	40	40
chip-all-clusters-app.elf	.flash.rodata	-24	-24

Full report output

BLOAT REPORT

Files found only in the build output:
    report.csv

Comparing ./master_artifact/chip-all-clusters-app.elf and ./pull_artifact/chip-all-clusters-app.elf:

sections,vmsize,filesize
.debug_loc,0,1653
.debug_info,0,1213
.debug_line,0,759
.debug_str,0,335
.strtab,0,171
.flash.text,40,40
.debug_frame,0,24
[Unmapped],0,24
.symtab,0,16
.debug_abbrev,0,14
.debug_aranges,0,8
.xt.prop._ZTVN4chip11DeviceLayer37DeviceNetworkProvisioningDelegateImplE,0,2
.shstrtab,0,-3
.flash.rodata,-24,-24
.debug_ranges,0,-48

[github-actions]

Size increase report for "nrfconnect-example-build" from cb23e0c

File	Section	File	VM
chip-lock.elf	text	44	44
chip-lock.elf	shell_root_cmds_sections	-12	-12
chip-lock.elf	rodata	-16	-20
chip-lighting.elf	text	40	40
chip-lighting.elf	shell_root_cmds_sections	-8	-8
chip-lighting.elf	rodata	-24	-20

Full report output

BLOAT REPORT

Files found only in the build output:
    report.csv

Comparing ./master_artifact/chip-lock.elf and ./pull_artifact/chip-lock.elf:

sections,vmsize,filesize
.debug_info,0,1126
.debug_loc,0,861
.debug_line,0,380
.debug_str,0,335
.strtab,0,171
.symtab,0,64
.debug_frame,0,60
text,44,44
.debug_ranges,0,16
.debug_abbrev,0,14
.debug_aranges,0,8
.shstrtab,0,1
shell_root_cmds_sections,-12,-12
rodata,-20,-16

Comparing ./master_artifact/chip-lighting.elf and ./pull_artifact/chip-lighting.elf:

sections,vmsize,filesize
.debug_info,0,1126
.debug_loc,0,865
.debug_line,0,384
.debug_str,0,335
.strtab,0,171
.symtab,0,64
.debug_frame,0,60
text,40,40
.debug_ranges,0,16
.debug_abbrev,0,14
.debug_aranges,0,8
.shstrtab,0,1
shell_root_cmds_sections,-8,-8
rodata,-20,-24

Comparing ./master_artifact/chip-shell.elf and ./pull_artifact/chip-shell.elf:

sections,vmsize,filesize

[andy31415]

Should we call these ResendMessage?

I think there should be a difference between stuff taking headers and things that decode packet/payload headers.

[pan-apple]

How about SendEncryptedMessage? From the API perspective, it does not know the CRMP is using it for resends.

[pan-apple]

@saurabhst, @jelderton, @BroderickCarlin, do you have any comments?