Remove CHIP buffer size check within shell and leave this it to lower layer

[yufengwangca]

Problem

What is being fixed?

We should not take Payload size check within the shell application since it does not take default header size into account,

For example, I set the buffer size to 1279, which passed the payload size check within Shell, but still failed with CHIP Error 4004

> send -s 1279 192.168.86.171
IP address: 192.168.86.171
CHIP:IN: TransportMgr initialized
CHIP:IN: TransportMgr initialized
CHIP:DL: writing settings to file (/tmp/chip_counters.ini.tmp)
CHIP:DL: failed to rename (/tmp/chip_counters.ini.tmp), Operation not permitted (1)
CHIP:IN: local node id is 0x000000000001b669
CHIP:IN: Connection from 'UDP:192.168.86.171:11097' expired
CHIP:IN: New pairing for device 0x0000000000bc5c01, key 0!!
Establish secure session succeeded

Send CHIP message with payload size: 1279 bytes to Node: 12344321
CHIP:EM: Failed to send message with err CHIP Error 4004 (0x00000FA4): Message too long
Send CHIP message failed, err: CHIP Error 4004 (0x00000FA4): Message too long
Send failed with error: CHIP Error 4004 (0x00000FA4): Message too long
Done
>

Change overview

1. Remove CHIP buffer size check within shell and leave this it to lower layer
2. Use stack buffer instead of dynamic memory to accommodate fixed length message.

Testing

How was this tested? (at least one bullet point required)
Set the Payload size to 1281 which exceed the previous kMaxPayloadSize

> ping -s 1281 192.168.86.171
IP address: 192.168.86.171
[1623341443.860683][150131] CHIP:IN: TransportMgr initialized
[1623341443.860804][150131] CHIP:IN: TransportMgr initialized
[1623341443.861217][150131] CHIP:DL: writing settings to file (/tmp/chip_counters.ini.tmp)
[1623341443.861499][150131] CHIP:DL: failed to rename (/tmp/chip_counters.ini.tmp), Operation not permitted (1)
[1623341443.861537][150131] CHIP:IN: local node id is 0x000000000001B669
[1623341443.861637][150131] CHIP:IN: Connection from 'UDP:192.168.86.171:11097' expired
[1623341443.861675][150131] CHIP:IN: New pairing for device 0x0000000000bc5c01, key 0!!
Establish secure session succeeded

Send echo request message with payload size: 1281 bytes to Node: 12344321
[1623341443.861834][150131] CHIP:EM: Failed to send message with err CHIP Error 4004 (0x00000FA4): Message too long
Send echo request failed, err: CHIP Error 4004 (0x00000FA4): Message too long
Send request failed: CHIP Error 4004 (0x00000FA4): Message too long
Ping failed with error: CHIP Error 4004 (0x00000FA4): Message too long
Done
>

Confirm the command send failed at lower layer due to "Message too long"

[bzbarsky-apple]

Much better, thank you! There's one remaining issue in SystemPacketBuffer and it's a little weird to be sending random

[bzbarsky-apple]

Please memset the part of the buffer up to the new data length to 0 or otherwise initialize it. Otherwise we are possibly leaking data that was in previous messages. Might be OK here, but in general it's a really bad pattern and we should not have it.

[yufengwangca]

Acked

[bzbarsky-apple]

Again, we should not be sending uninitialized data.

[yufengwangca]

Acked

[github-actions]

Size increase report for "nrfconnect-example-build" from 6f297c5

File	Section	File	VM
chip-shell.elf	device_handles	8	8
chip-shell.elf	rodata	-16	-20
chip-shell.elf	text	-56	-56

Full report output

BLOAT REPORT

Files found only in the build output:
    report.csv

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

sections,vmsize,filesize
.debug_line,0,83
.debug_info,0,65
.debug_ranges,0,48
device_handles,8,8
rodata,-20,-16
text,-56,-56
.debug_str,0,-107
.debug_loc,0,-361

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

sections,vmsize,filesize

[bzbarsky-apple]

Much better, thank you!