From 96ca7ee67a36605a33de8d57eb8e8d3998bc6427 Mon Sep 17 00:00:00 2001 From: Michał Masłowski Date: Wed, 03 Sep 2014 14:30:21 -0400 Subject: Merge libreboot-6b6. Conflicts: buildrom-withgrub --- (limited to 'docs/future/gnutoo_fallback_patch') diff --git a/docs/future/gnutoo_fallback_patch b/docs/future/gnutoo_fallback_patch new file mode 100644 index 0000000..50af779 --- /dev/null +++ b/docs/future/gnutoo_fallback_patch @@ -0,0 +1,182 @@ + I documented it + but I should update the page + somehow it works without any but one of my extra patches + but it has 1 small issue + GNUtoo-irssi: do you need review on those? I think that sth like it could save me countless external reflashs + phcoder-1creen: well, most of them are unnecessary now + 1 patch is usefull only for improving code readability of existing coreboot code + 1 patch is only changing the reboot count of the fallback mecanism + beside that I see nothing remaining + but I can check again + I have to do it now + The documentation is on the wiki + caveats: + 1) sometimes the x60 reboots twice, + for instance if you run poweroff, then let it power down, and as soon as it seems powered down, you press the power button + in that case it will do a reset + 2) suspend/resume and userspace needs some handling, I've systemd units for booting only, but not for suspend/resume + but you can do it by hand + config MAX_REBOOT_CNT + int + default 1 + that's what I added in src/mainboard/lenovo/x60/Kconfig + before I had a patch to make it selectable it in Kconfig, + that is to say the user enter the max reboot count he wants + I think the global default is 3 + Then I've some other interesting patches + I wonder if they're acceptable + one patch is for adding etc/grub.cfg from Kconfig + Use case: the user builds once, he do ./build/cbfstool ./build/coreboot.rom add -n etc/grub.cfg -f grub.cfg -t raw + but he re-do make + and forgett to re-add grub.cfg + it's just a convenience + (he could do it with a script too) + *he/she + I guess the user is a she in english? + en french it's a he + I've also a flashrom patch to submit + phcoder-1creen: "it could save me countless external reflashs" => that was exactly my use case + There are some other interesting stuff that could extend the use case: + there is a flash log for the chromebooks + example use case: you go to a conference in the USA, you are in the plane + you then continue developing there, you reflash etc... + but then you need the log of the failed boot somehow + the flash log (which is in coreboot but require CONFIG_CHROMEOS or something like that) could help with that second use case + Else the logs in RAM + a watchdog could also do the trick + *hardware watchdog + so that second approach of the second use case would just require some modifications related to cbmem + they may already be there, because I'm way out of the loop + I'll make a list of the interesting patches I have locally + and look at gerrit too + btw, is there some easy infrastructure work to do? + like something that can be done on the side +* ttyS3 has quit (Ping timeout: 264 seconds) + The x60[s/t], T60(with intel GPUs), are mostly complete, the main issue remaining is merging that improved GPU init code + fallback/ is mostly merged but that one patch I was talking about + then I guess the ACPI part was merged + I'm unsure about the IRDA + I mostly test on x60t nowadays + (my t60 has a nasty bug with ctrl+d, probably ec related) + I've also to look about the security of the I/Os + (like what's on the dock connector) + there is also the license issue of the microcodes inside the headers + I'll add all that in the wiki + GNUtoo-irssi: did you test digitizer? + yes + works well with libreboot 6 beta3 patches on top of coreboot git + I use it often + with xournal mainly + I've been in a local shop and I've found a compatilble wacom pen: it has: + touch, button(right click), eraser + all do work + the pen is not the x60 pen, but it does work fine + digitizer patches are already in + The screen's directional keys the its middle key work + yes + I'll update soon + I'll probably sumarize the patch I've left in the wiki + and update that fallback page + phcoder-1creen: is the IRDA supposed to work? + GNUtoo-irssi: 5243 + T60, rght? +* KidBeta has quit (Quit: My MacBook Pro has gone to sleep. ZZZzzz…) + x60 and x60t + oops + x60t and t60 + I'll test them together, I was trying lirc instead directlyt + 5242 for X60 + ok + thanks + GNUtoo-irssi: did you see x200 port? + yes + what CPU is it? + and what chipset? + gm45. Intel GPU + ok + I've looked at your new ports and related, + it probably cover the chipset I have in my N71JQ + but I probably don't have time to do the port anytime soon + gm45 was already covered by rk9 + yes + it's core 2 duo with the first AMT in the NICs, right? + and 64bit? + it's 64bit. I can't tell anything about AMT. + so I guess that if someone unsolder his nic firmware flash, the AMT is gone? + ok + if so that's probably a good tradeoff + you get more recent laptops at the cost of unsoldering or blanking the NIC's flash + ME firmware is in the flash chip. There is information that on gm45 you can remove ME firmware without any consequences but I din't really try + assuming it's like with the old i945 laptops + ok, wow, nice + how fast is it in between the T60's and the Nehalem's laptops(x201) + roda rk9 runs without ME firmware AFACIT + ok + about roda and so on, there isn't a lot of infos on the rugged laptops + I guess that nobody still test on them + No. But the list of connectors they have is truly impressive. As is battery capacity and heaviness. + indeed + it probably has lot of interesting peripherals too, like GPS, 3g modem(how is it connected?) and so on + for the heavyness, it's a way to make geeks become like rambo? + s/geeks/geeks and nerds + 3g modems are optional. I guess it's minipcie slot. + BTW x200 has 3 minipcie slots + wow + (not counting exprecsscard) + ok + that permits to have 2 wifi cards... + if driver can handle it, sure. When I tried with 2 intel cards, intel drivers and networkmanager got confused. + (ath9k/ath5k have some difficulties when creating multiples interfaces when WPA is involved) + ok + 3rd minipcie was intended for UWB. + well, I have multiples cards easily here + I never had a problem with non-intel cards + network manager will still get confused + example: ath9k + ath9k_htc => both interfaces appear in kde's network manager GUI + it was getting confused with intel cards and rfkill + (and I lacked the fimrware of the intel cards...so that added to the confusion) + Example use case: connect to 2 different AP on 2 different networks + yes network manager and multiple cards and rfkill resultsin confusion + my ath9k_htc is usb + so no hardware rfkill + btw, the mini-pcie connectors do export only pci? + do they export usb, and sata? + (and some other pins for rfkill, SIM card, and so on) + + + + + + + + ok + well, I must update the instructions + I was going trough the list of patches I had first + yes + but to a specific/personal page + could you link me to the updated instructions? (when done) + well, I'll update them first + I was going trough my patches list before that + so I'll do that now + So I gather that you basically reset the counter yourself after you boot (after typing grub password) + and so, if you boot and the counter is higher, you know if someone tried to use it + yes, my systemd unit does it + *resets it + so it works like that: + the bootblock switch from normal/ to fallback if the counter is > CONFIG_MAX_REBOOT_CNT + if no normal/ is there it also switch to fallback/ + and then it increments the counter + (it's badly explained by me but you get the idea) + then my systemd units reset the counter to 0 once it's fully booted + that way if it fails, let's say at booting any linux kernel, then the user won't have bricked the laptop + (and the developer will have saved lot of time) + the issue is that I didn't reset the counter at resume + I should look how + but at least that makes it developer friendly if the user don't have suspend-resume covered yet + testing images is then a lot faster + and for "production", only fallback/ populated, but with the mecanism in place + that way he can test normal/ easily + + + + + -- cgit v0.9.1