Hi Wolfgang,
On Wed, Mar 7, 2012 at 10:30 PM, Wolfgang Denk wd@denx.de wrote:
Dear Lei Wen,
In message CALZhoSSzBhmZ_c79zKBQE7Z5YEtCKC_E-4Vgr35rmBLiV_LawQ@mail.gmail.com you wrote:
I miss add the zconf.h into the patch... V2 patch is attached...
Argh. Please NEVER send patches as attachments. Always send them inline. See http://www.denx.de/wiki/U-Boot/Patches
Sorry for that again. I send the patch as attachment, which is only for RFC, and another reason is the patch itself is very huge and almost all content is kept as it is from zlib main stream. Anyway, I would keep the send patch rule in mind next time.
As it copy from zlib home page, the improvement from 1.2.5 to 1.2.6 contains following:
gzread() can now read a file that is being written concurrently gzgetc() is now a macro for increased speed Added a 'T' option to gzopen() for transparent writing (no compression) Added deflatePending() to return the amount of pending output Allow deflateSetDictionary() and inflateSetDictionary() at any time in raw mode deflatePrime() can now insert bits in the middle of the stream ./configure now creates a configure.log file with all of the results Added a ./configure --solo option to compile zlib with no dependency on any libraries Fixed a problem with large file support macros Fixed a bug in contrib/puff Many portability improvements
This is a lot of changes / improvements for genral use, but which of these are actually useful for U-Boot?
I. e. what exactly is the motivation to switch to a new version of the code? Do you want to fix any specific bugs, or improve performance, or what?
What I see is that the code size grows by about 2 kB (measured on ARM):
-> bloat-o-meter u-boot-before u-boot-after add/remove: 23/2 grow/shrink: 2/5 up/down: 4523/-2496 (2027) function old new delta static.lenfix - 2048 +2048 inflateCopy - 352 +352 updatewindow - 308 +308 inflateSync - 276 +276 inflateSetDictionary - 192 +192 static.adler32_combine_ - 156 +156 inflateReset2 - 152 +152 inflateResetKeep - 148 +148 static.distfix - 128 +128 inflatePrime - 116 +116 inflateMark - 104 +104 syncsearch - 84 +84 zmemcmp - 68 +68 inflateSyncPoint - 64 +64 inflateGetHeader - 64 +64 inflate_copyright - 47 +47 inflateUndermine - 44 +44 inflate 5764 5800 +36 zmemzero - 28 +28 zmemcpy - 28 +28 zError - 24 +24 adler32_combine64 - 20 +20 adler32_combine - 20 +20 zlibVersion - 12 +12 zunzip 264 268 +4 inflateEnd 108 100 -8 inflateInit2_ 264 208 -56 inflate_table 1492 1428 -64 inflateReset 128 48 -80 inflate_fast 1480 1368 -112 distfix 128 - -128 lenfix 2048 - -2048
I wonder which benefits we get for this price we are paying?
The main reason I'd like to introduce this upgrade is for I want to add the compressing feature for uboot. And the 1.2.6 has some fix for the deflate, so it is maybe a good base line for introducing it.
Also the reason why I want to enable the compressing is that sometimes we may want to dump whole ram in the uboot, and since the ram grows bigger and bigger, compressing and then send to host is a reasonable method to reduce the dumping time.
How about define a Macro like CONIFG_ENABLE_GZIP_COMPRESSION to compile the compression related code only when this flag is on?
Best regards,
Wolfgang Denk
Thanks, Lei