00:39:40 | Sammy | morning... |
07:38:50 | Sammy | Amm.. if I what my LCD with Touch panel , is that any driver can use in kernel ? |
07:39:19 | Sammy | about the touch panel ... |
07:41:16 | Sammy | and is that any effect about App ? if the driver is good or bad ? |
11:03:40 | erikm | hi |
11:03:57 | Sammy | evering erikm |
11:08:26 | Sammy | erikm: someday before I asked you that you say LCD driver use the kernel sa1100fb.c ... |
11:09:16 | Sammy | and is that any driver about Touch panel in kernel too ? |
11:18:32 | erikm | there is a UCB1200 touch screen driver, don't know in what state it currently is |
11:23:57 | Sammy | before I see the MTD and jffs How to , what's the different between jffs and RAMdisk , sound's like the same thing ,right ? |
11:25:49 | erikm | ramdisk lives in ram. JFFS2 filesystem lives in flash |
11:27:18 | Sammy | but ramdisk compress file also in flash ...? |
11:28:14 | erikm | no, compressed ramdisk lives in flash, and is decompressed into ram |
11:28:20 | erikm | that's why it is called a ramdisk |
11:28:32 | Sammy | what's jffs2 actually doing ? |
11:28:40 | erikm | JFFS2 lives in flash, and if you write to it, your flash will be updated |
11:29:38 | Sammy | how to write ? |
11:30:32 | erikm | same way as you write to a normal disk. fd=open(file); write(fd, buf, len); close(fd); |
11:30:51 | erikm | MTD+JFFS2 make a real block device from the flash |
11:33:20 | Sammy | Amm..if I copy a file from zmodem into ram then use MTD+jffs2 , and it's can updated into flash soon ? right ? |
11:34:00 | erikm | yes. so next time you'll reboot the file is still there |
11:35:08 | Sammy | erikm: something I still get clear , about blob and kernel ,ramdisk catch LART 4MB flash . |
11:35:40 | Sammy | but if I want to run QT and some app , where can I put in flash ? |
11:36:07 | erikm | not. |
11:36:19 | Sammy | ? |
11:36:37 | Sammy | can't ? or I think wrong ? |
11:36:43 | erikm | it is quite easy to develop a board for the high speed bus that adds more flash to the second static memory bank |
11:36:56 | erikm | can't |
11:37:31 | erikm | we also have a PCMCIA board in development that allows you to add compact flash modules |
11:37:36 | erikm | or an IBM microdrive |
11:38:39 | Sammy | it's means that all QT and App need to put in something like CF card ... |
11:39:52 | erikm | yes |
11:40:33 | Sammy | is that sa-1100 support 640*480 LCD panel or it can support to 800*600 LCD panel ? |
11:41:44 | erikm | it supports anything up to 1024x1024 *as long as it fits in the memory bandwidth* |
11:45:27 | Sammy | but what frequency for sa-1100 use ? 190 or 221 ? because when we are calc the clock for LCD mode line it's not enough ?why ? |
11:56:07 | erikm | the sa1100 on the LART is spec'ed at 221MHz |
11:56:56 | erikm | it's not the clock |
11:57:20 | erikm | it's the memory bandwidth requirement of the LCD DMA controller |
11:57:41 | erikm | Intel has an application note about that on their strongarm pages |
12:03:45 | Sammy | but from the sa-1100 develop Manual say that PCD(pixel Clock divisor) ,pixel Clock Frequency = CCLK/2(PCD+2) , and when CCLK = 221 , but 800*600 can accept the freq is 38MHZ ~ 48MHZ , |
12:04:44 | Sammy | and it's not in the side ... |
12:07:36 | Sammy | not in bandwidth support ...why is this ? confuse ? |
12:08:01 | erikm | it is easy |
12:08:10 | erikm | suppose the memory bandwidth is 100MB/s |
12:09:00 | erikm | suppose you want to do 1024x768, 16bpp, 60Hz |
12:09:22 | erikm | so you need 1024*768*2*60 = 90MB/s |
12:09:40 | erikm | that won't fit |
12:09:59 | Sammy | no , 800*600 ,16bpp ,38~48Mhz |
12:10:14 | erikm | because you also need some headroom for the DRAM refresh cycles and the CPU itself |
12:10:53 | erikm | what's the refresh frequency for that mode? |
12:11:21 | Sammy | 70 HZ |
12:12:50 | erikm | that needs 64MB/s |
12:13:23 | erikm | we already got problems with 640x480 16bpp 60Hz |
12:13:34 | erikm | so you better go for an 8 bit mode |
12:18:01 | Sammy | who to calc that ? is that ( 800*600*16*70) / (1024*1024) ? |
12:18:16 | Sammy | er how ? |
12:20:37 | Sammy | sorry 800*600*2*70 = 64 MB right ? |
12:24:48 | Sammy | but turelly we want to use UNIPAC UB104S01 800*600 color TFT-LCD panel ,how can we currect the MB/s into 38~48 MHz ? |
12:25:07 | erikm | yes |
12:25:16 | erikm | 800x600x2x70 |
12:26:31 | erikm | that's more than half the memory bandwidth |
12:26:49 | erikm | (which is about 110MB/s according to lmbench) |
12:39:06 | erikm | spots an obvious bug in src/serial.c |
12:40:14 | Sammy | great ... |
12:51:15 | Sammy | 's time to go home ... |
12:51:36 | Sammy | night erikm , and.....................Thanx |
12:54:23 | prpplague | morning all |
12:56:18 | Sammy | night. |
12:56:57 | prpplague | that too... |
13:06:28 | erikm | morning prpplague |
13:10:52 | prpplague | working hard today, or hardly working today? |
13:11:13 | erikm | doing five things at the same time |
13:12:19 | prpplague | i know what you mean... |
13:15:31 | erikm | sheesh, memory is really cheap |
13:16:10 | erikm | just got 1GB of memory. 4 256MB DIMMs, about 40 US$ each. |
13:19:48 | ed__ | rock on |
13:19:53 | prpplague | i'm ready for flash to be that cheap |
13:21:24 | ed__ | that ain't gonna happen |
13:21:26 | ed__ | :P |
13:22:47 | prpplague | can dream |
13:23:06 | ed__ | no i think they passed laws against that |
13:25:56 | erikm | ed__: of course, that's the US, right? :) |
13:27:33 | prpplague | a law? guess i've missed that, what was the reason behind a law? |
14:10:57 | prpplague | erikm: if the sa-1100 device i'm working with has a single 16meg ram bank would place the initrd at 0xc0400000 or 0xc0800000? russ tells me that it really doesn't matter where you put the initrd image as long as you have room... |
14:12:58 | erikm | he's right. the kernel will get rid of the initrd image as soon as it's uncompressed |
14:13:20 | erikm | the name ATAG_INITRD is misleading |
14:13:33 | erikm | it's actually ATAG_COMPRESSED_RAMDISK_IMAGE |
14:16:01 | prpplague | the kernel will read from the 0xc0000100 tags where the initrd image is located in ram and pass that information on to the ramdisk driver, correct? |
14:16:12 | erikm | yes |
14:16:20 | erikm | and the ramdisk driver will decompress it |
14:16:27 | prpplague | cool, its all starting to make sense... |
14:17:06 | erikm | and because we tell it params->u.ramdisk.start = 0, it will decompress it at whatever position it likes |
14:23:11 | prpplague | i've notice that there are some differences in how some platforms are handling tags in the fixup sections |
14:25:21 | erikm | nod |
14:25:43 | erikm | in an ideal world you don't want any stuff in the fixup code |
14:26:28 | prpplague | the bootloader should pass the correct info to the tag ram address and the kernel should read the info from there |
14:26:46 | erikm | yes |
14:31:30 | prpplague | gotta ya |
14:32:07 | prpplague | continues the exploratory surgery of blob |
16:00:33 | erikm | re, guys |
16:01:30 | erikm | Stefan Eletzhofer has memory test routines |
16:02:31 | erikm | I asked him to use the latest CVS sources so he can use the new command line parser to add them |
16:03:05 | prpplague | cool, when do expect a comit? |
16:03:12 | erikm | I don't know yet |
16:04:17 | erikm | he doesn't have CVS write access yet, I first want to see if his solution looks right |
16:07:32 | BZFlag | have him send the patch into the SF patches database. |
16:07:47 | erikm | BZFlag: good idea, in that way I also learn how to use it :) |
16:11:34 | BZFlag | he does a cvs diff -u > foo.patch and then opens a patch and attaches the file. then anyone can test it. |
16:13:12 | erikm | BZFlag: ok, thanks |
16:20:11 | prpplague | is away: lunch |
16:21:05 | phil_trick | is away: I'm busy |
16:24:22 | prpplague | is back (gone 00:04:10) |
16:25:16 | prpplague | well then i guess i'm not going to lunch |
16:26:25 | erikm | prpplague: -ETANSTAAFL? :) |
16:27:20 | prpplague | erikm: ok i give it what is that? |
16:27:47 | erikm | TANSTAAFL: There Ain't No Such Thing As A Free Lunch |
16:28:09 | prpplague | lol, thats for sure |
16:28:16 | erikm | or as a coworker always says: free cheese only comes in mouse traps |
16:28:25 | prpplague | like that one better |
20:39:34 | prpplague | howdy russ |
20:54:02 | erikm | goes zzz |
21:28:30 | prpplague | night all |