| 00:00.44 | Kevin2 | cr2/donn: The "addwatch" command is now "addlist gpios". |
| 00:02.56 | pH5 | hi kevin, I think I understand most of it now. the part where I'm uncertain is what should I write to the page table. |
| 00:03.32 | pH5 | I know it's at phys a0090000 - are there some docs for MMU newbies about its format? |
| 00:03.36 | Kevin2 | pH5: Okay - the trampoline is at virt 92000000 and phys a2000000. Haret will start running mmu_trampoline at virt 92000000, and then change the mmu for virt a2000000 to point to a2000000 physical (see the comment "Turn off mapping for this code"), and then jump to virt a2000000. Once running at an address that is both the same virt/phys it turns off the mmu. |
| 00:04.24 | Kevin2 | pH5: What is it you are trying to do? |
| 00:05.12 | pH5 | I'm injecting code into the running htc bootloader (appending, rather). The goal is to boot linux from it. |
| 00:06.36 | Kevin2 | Isn't the mmu off already in the boot loader? |
| 00:07.37 | Kevin2 | For mmu docs see the ARM reference manual: http://www.altera.com/literature/third-party/ddi0100e_arm_arm.pdf |
| 00:08.57 | pH5 | thanks. the htc loader copies itself (or parts) into ram and turns on the mmu early. |
| 00:11.08 | Kevin2 | pH5: That is a pain. Do you know if they use 4K pages or 1Meg mappings? 1Meg mappings will be easier for you to deal with. |
| 00:13.18 | Kevin2 | BTW, there is a "resume into boot" feature in haret. (cr2 mentioned this earlier.) Unfortunately, it does not work. |
| 00:13.23 | pH5 | it's a pxa272, I think they always use 1m mappings |
| 00:13.55 | Kevin2 | The idea is to hook the bootloader's resume vector to launch linux. I think the bootloader has caching enabled which causes the boot to fail. However, I haven't yet confirmed this. |
| 00:14.38 | pH5 | but at least on wm2003 I think parts of the kernel are executed in place from flash, I want to work around wince altogether. |
| 00:16.12 | Kevin2 | I think a number of other projects got the bootloader to launch linux. |
| 00:20.03 | regulate | anyone else have a hermes? |
| 00:23.47 | Kevin2 | regulate: Yes, I do. |
| 00:25.10 | regulate | is there a newer kernel and initram than available here http://handhelds.org/~koconnor/HTCHermes/ ? |
| 00:26.28 | Kevin2 | No, but I can upload a new one if you like. |
| 00:26.28 | regulate | i've tried building one from svn, using htchermes_defconfig |
| 00:26.35 | regulate | it wouldn't boot however |
| 00:28.03 | Kevin2 | regulate: I used an irq patch in my kernel. |
| 00:29.14 | regulate | can i extract the .config from that zImage ? |
| 00:31.36 | Kevin2 | regulate: http://handhelds.org/~koconnor/HTCHermes/config-20070607 |
| 00:31.47 | pH5 | hm... so the mapping for virt 0x92000000 is at page_table+0x2480 ... |
| 00:31.52 | Kevin2 | there is also: http://handhelds.org/~koconnor/HTCHermes/herm-linux-20070607.exe |
| 00:31.57 | pH5 | Kevin2: thanks again, I should have read that much earlier :) |
| 00:32.20 | pH5 | Kevin2: how far along is hermes by now? |
| 00:33.14 | Kevin2 | pH5: It boots, but not much has been done with driver writing. cr2 has mapped out a whole bunch of things, but no real progress on the implementation side. We really need a storage device (nand or sd) to boot linux for real. |
| 00:36.35 | Kevin2 | regulate: The irq patch is at: http://handhelds.org/~koconnor/HTCHermes/herm-diff-20070607 |
| 00:36.53 | Kevin2 | You should be able to boot from CVS with that patch applied. |
| 00:44.46 | regulate | hmm |
| 00:44.53 | regulate | can't compile with EABI option |
| 00:45.03 | regulate | cc1: error: invalid option `abi=aapcs-linux' |
| 00:45.42 | regulate | make ARCH=arm CROSS_COMPILE=/usr/local/arm/3.4.1/bin/arm-linux- |
| 00:46.36 | Kevin2 | regulate: I suggest you get a newer arm cross compiler. EABI is pretty common now. If you don't want to bother, just turn that off in the config. |
| 00:48.47 | regulate | making without the option first |
| 00:49.07 | regulate | i can't find the deb for arm-linux-gcc |
| 00:54.38 | regulate | Kevin2: using herm-linux*.exe , how is usb0 configured? |
| 00:56.05 | Kevin2 | 192.168.0.206 |
| 00:56.34 | Kevin2 | You need to have usb0 plugged in during the boot. After the boot completes, unplug and then replug in the usb cable. |
| 00:59.14 | pH5 | good night |
| 00:59.22 | regulate | how do i connect to it from my workstation |
| 00:59.38 | regulate | <PROTECTED> |
| 01:00.24 | Kevin2 | telnet 192.168.0.206 |
| 01:01.13 | Kevin2 | regulate: Did you bring up the usb0 interface on your host? (eg, ifconfig usb0 192.168.0.200) |
| 01:01.25 | regulate | yeah i just did that |
| 01:03.11 | Kevin2 | Can you telnet to the device? ping it? |
| 01:03.15 | regulate | yup |
| 01:15.17 | regulate | ok, rebuilt using your config |
| 01:16.28 | regulate | minus the EABI option. using your rootfs. |
| 01:16.59 | regulate | init reports cannot create /sys/class/graphics/con_rotate |
| 01:18.12 | Kevin2 | No big deal - the rootfs was reused from a different platform and there is some junk in there. |
| 01:18.29 | regulate | i have trouble extracting it |
| 01:18.50 | regulate | cpio wants to put things in / |
| 01:19.41 | Kevin2 | Hrmm. Not sure why that is. |
| 01:20.59 | regulate | files in the archive start with / |
| 01:23.22 | Kevin2 | Worked for me: gunzip < ../rootfs.cgz | cpio -i |
| 01:24.06 | Kevin2 | Why do you want to extract it? You can't really put it back together again (it needs to create special device files). |
| 01:25.00 | regulate | regulate@dab:~/code/htc/temp$ gunzip < ../rootfs.cgz | cpio -i |
| 01:25.00 | regulate | cpio: /dev: Cannot change mode to rwxr-xr-x: Operation not permitted |
| 01:25.04 | regulate | thats what i get |
| 01:25.59 | Kevin2 | $ cpio --version |
| 01:25.59 | Kevin2 | cpio (GNU cpio) 2.6 |
| 01:26.48 | regulate | 2.7 |
| 01:28.25 | regulate | hmm can i mount this image? |
| 01:28.37 | Kevin2 | No |
| 01:29.06 | regulate | Kevin2: http://wiki.xda-developers.com/index.php?pagename=HowToCompile4Himalaya |
| 01:29.09 | regulate | i was reading that |
| 01:29.24 | regulate | towards the bottom are instructions for rebuilding an initrd |
| 01:33.50 | Kevin2 | I'm using an initramfs which is different from initrd. You can mount initrd, but you can't mount an initramfs. It is generally easier to work with initramfs, however. |
| 01:43.40 | regulate | do you have any instructions for creating one? |
| 02:07.21 | regulate | do you use gen_init_cpio ? |
| 02:07.52 | Kevin2 | regulate: I have some old directions which I just uploaded: http://handhelds.org/~koconnor/initramfs/ |
| 02:09.21 | Kevin2 | As you suspected, I now call gen_init_cpio directly instead of building the initramfs directly into the kernel |
| 02:09.40 | Kevin2 | ../hh-head-2.6.17/usr/gen_init_cpio filelist | gzip > rootfs.cgz |
| 02:14.00 | regulate | ok |
| 02:21.58 | Kevin2 | Good night. |
| 02:41.24 | *** join/#htc-linux mndctrl (n=asbjorn@2.81-167-1.customer.lyse.net) |
| 05:00.55 | *** join/#htc-linux rmoravcik (n=rmoravci@pc-3s0zt5w2e4y0vzmhnrzq3a21zqajzfw.users.student.utc.sk) |
| 05:31.10 | *** join/#htc-linux sadeness_ (n=LamersIn@nat2.rlan.ru) |
| 05:35.10 | *** join/#htc-linux RoEn (n=roen@p54a65fb6.dip.t-dialin.net) |
| 05:47.00 | *** join/#htc-linux don (n=don@p5085C1C3.dip.t-dialin.net) |
| 07:01.23 | *** join/#htc-linux dion (n=dion@inhex.net) |
| 07:26.06 | *** join/#htc-linux rob__w (n=bob@BACa9df.bac.pppool.de) |
| 08:22.26 | *** join/#htc-linux pH5 (n=ph5@e178223143.adsl.alicedsl.de) |
| 08:35.58 | *** join/#htc-linux don (n=don@p5085c1c3.dip.t-dialin.net) |
| 08:44.58 | *** join/#htc-linux psokolovsky (n=psokolov@empty-crossing.volia.net) |
| 09:13.17 | *** join/#htc-linux BabelO (n=Miranda@faim1.itesoft.com) |
| 09:13.33 | BabelO | hi |
| 09:15.09 | BabelO | psokolovsky : i m writing cs driver for acx, i use CIS read to setup correctly pcmcia. got issue with tuple actually :) http://pcmcia-cs.sourceforge.net/ftp/doc/PCMCIA-PROG-3.html |
| 09:15.55 | psokolovsky | Hi! |
| 09:15.58 | psokolovsky | BabelO: good news! |
| 09:16.40 | BabelO | psokolovsky: when you find correct value for COR and SCR, you read it from CIS dump ? |
| 09:17.47 | psokolovsky | BabelO: well, I didn't find anything ;-) |
| 09:19.12 | BabelO | i think i need you to reimplement driver correctly, actually i duplicate mem.c to cs.c and then add specifics cs function. i also take your h4000_pcmcia as sample but did not understand how your pcmcia power line can be shutdown ? |
| 09:24.46 | *** join/#htc-linux DTMonterrey (n=evandro@cornelius.estig.ipb.pt) |
| 09:46.40 | psokolovsky | BabelO: well, I can say what I said - I'll be glad to help, but dunno when I'll have enough time ;-( |
| 09:46.58 | psokolovsky | anyway, if you'll have something viable, I guess I'll find time ;-) |
| 09:55.24 | BabelO | psokolovsky: ok i think i send you something this evening or in weekend. i duplicate mem.c to cs.c because i don't want to remove static declaration . |
| 09:55.24 | psokolovsky | ok |
| 11:31.02 | *** join/#htc-linux donn (n=donn@80.133.193.195) |
| 11:32.24 | donn | hi cr2 |
| 11:32.29 | donn | hi kevin2? are you here? |
| 11:43.16 | *** join/#htc-linux sadeness (n=LamersIn@nat2.rlan.ru) |
| 11:43.30 | pH5 | hmmh, I can see the linux decompressor fill the framebuffer now. seems I managed to jump into the kernel. |
| 11:56.16 | *** join/#htc-linux pleemans (n=peter@d5152D19B.access.telenet.be) |
| 12:02.48 | *** join/#htc-linux pleemans (n=peter@d5152D19B.access.telenet.be) |
| 12:09.22 | *** join/#htc-linux pleemans (n=peter@d5152D19B.access.telenet.be) |
| 12:29.57 | *** join/#htc-linux psokolovsky_ (n=psokolov@empty-crossing.volia.net) |
| 12:52.44 | *** join/#htc-linux BabelO (n=Miranda@faim1.itesoft.com) |