refactor rcs_chunked_list and remove its c++ features

[jiangzidong]

remove rcs_chunked_list 's c++ features. All the functionality remain the same

It can pass the make precommit.

in x86 machine (32bit)
size -A release_jerry output

section	master	patch
.note.gnu.build-id	36	36
.text	253351	1253191
.rodata	13564	13596
.eh_frame	26464	26036
.init_array	4	0
.data	1096	1096
.bss	2363776	2363776
.comment	41	41
Total	2658332	2657772

[jiangzidong]

performance in x86 machine. remain same

Benchmark	RSS (+ is better)	Perf (+ is better)
3d-cube.js	116-> 116 (0.000)	0.284-> 0.288 (-1.408)
access-binary-trees.js	84-> 84 (0.000)	0.16-> 0.164 (-2.500)
access-fannkuch.js	44-> 44 (0.000)	0.836->0.8368 (-0.096)
bitops-3bit-bits-in-byte.js	32-> 32 (0.000)	0.236-> 0.236 (0.000)
bitops-bits-in-byte.js	32-> 32 (0.000)	0.2936-> 0.288 (1.907)
bitops-bitwise-and.js	36-> 36 (0.000)	0.292-> 0.292 (0.000)
controlflow-recursive.js	240-> 240 (0.000)	0.144-> 0.14 (2.778)
crypto-aes.js	124-> 124 (0.000)	0.528->0.5296 (-0.303)
crypto-md5.js	188-> 188 (0.000)	2.4352-> 2.464 (-1.183)
crypto-sha1.js	136-> 136 (0.000)	1.124-> 1.136 (-1.068)
date-format-tofte.js	76-> 76 (0.000)	0.296-> 0.292 (1.351)
date-format-xparb.js	72-> 72 (0.000)	0.156-> 0.148 (5.128)
math-cordic.js	40-> 40 (0.000)	0.32-> 0.316 (1.250)
math-spectral-norm.js	40-> 40 (0.000)	0.192-> 0.192 (0.000)
string-base64.js	168-> 168 (0.000)	28.6832->29.3656 (-2.379)
string-fasta.js	52-> 52 (0.000)	0.5776->0.5832 (-0.970)
Geometric mean:	RSS reduction: 0%	Speed up: 0.1752%

[rtakacs]

Could you please repeat these measurements on an ARM board? (The best would be on an RPI2)

[rtakacs]

I've already started work on this topic (#791 (comment)) and I have a not-yet-published patch very similar to yours, but I intended to keep it private until all the code size increase issues I've experienced are solved.

To reduce the chance of duplicating work: I've also been working on the refactoring of the operand class in the parser.

[zherczeg]

Yes I agree to avoid work duplication. I think you should discuss how to share the work. Perhaps a good topic for Gitter :)

[jiangzidong]

measure on RPI2, @rtakacs is the data similar with yours?

Benchmark	RSS (+ is better)	Perf (+ is better)
3d-cube.js	112-> 112 (0.000)	2.97-> 2.99 (-0.673)
access-binary-trees.js	84-> 80 (4.762)	1.84667-> 1.82 (1.444)
access-fannkuch.js	40-> 36 (10.000)	8.67->9.00333 (-3.845)
bitops-3bit-bits-in-byte.js	32-> 32 (0.000)	2.22333->2.23667 (-0.600)
bitops-bits-in-byte.js	32-> 32 (0.000)	2.99-> 2.99 (0.000)
bitops-bitwise-and.js	32-> 32 (0.000)	3.44-> 3.61 (-4.942)
controlflow-recursive.js	220-> 220 (0.000)	1.53667->1.55667 (-1.302)
crypto-aes.js	124-> 120 (3.226)	5.14667->5.21667 (-1.360)
crypto-md5.js	184-> 184 (0.000)	24.3367->24.4133 (-0.315)
crypto-sha1.js	132-> 136 (-3.030)	11.22->11.1833 (0.327)
date-format-tofte.js	72-> 72 (0.000)	3.19333->3.21333 (-0.626)
date-format-xparb.js	76-> 80 (-5.263)	1.65667->1.63667 (1.207)
math-cordic.js	40-> 36 (10.000)	3.27-> 3.29 (-0.612)
math-spectral-norm.js	40-> 36 (10.000)	2.05333-> 2.05 (0.162)
string-fasta.js	48-> 48 (0.000)	5.29333->5.37667 (-1.574)
Geometric mean:	RSS reduction: 1.9588%	Speed up: -0.834%

section	master	patch
.note.gnu.build-id	36	36
.text	189192	188320
.rodata	8284	8336
.ARM.exidx	8	8
.init_array	4	0
.data	1096	1096
.bss	2363776	2363776
.comment	43	43
.ARM.attributes	51	51
Total	2562490	2561666

binary size:
master 198K
patch 194K

[zherczeg]

So this is actually a size reduction! Great!

[jiangzidong]

@zherczeg Do you mean the RSS result? I don't know why the rss will change in RPI2. In x86 machine, it is exactly same after the refactor.

[rtakacs]

@jiangzidong: My last measurement with my patch was at the end of December. The code size was a bit bigger than yours, but the performance test results were similar to yours.

[ruben-ayrapetyan]

Usually, we place comment with function name after closing bracket, related to the function.
In the case, the line should be the following: } /* rcs_chunked_list_set_prev */.

[jiangzidong]

@ruben-ayrapetyan Thanks, I fixed as you commented.

[ruben-ayrapetyan]

Looks good to me

[rtakacs]

LGTM, but I'm not a reviewer.

[zherczeg]

LGTM

[rtakacs]

@jiangzidong: I'm compiling the project with a standard C compiler and fixing the remaining little C++ features. Before we land this patch, please add a 'typedef' to the rcs_chunked_list_t struct to use it as a type.

[jiangzidong]

@rtakacs: updated. sorry for the mistake

[rtakacs]

@jiangzidong Thanks for the update. The fix looks good to me.

[LaszloLango]

Landed (dfb22c3)