AnOldHacker | gnarface onefang fsmithred rwp Thanks for your advice. In the end, I managed to clean things up from the live thumbdrive installation. Since I'm not a regular, I figure it is my duty to inform as best as I can. I'll try to hang around, so if there is a better place to document, please ping me. Situation: Dell Latitude 7520. BIOS version 1.24.1. i7-1185G7/4.8GHz. Came installed with Windows. Tried to go multiboot w/out reading i | 01:37 |
---|---|---|
AnOldHacker | tructions. Blew away EFI partition. Tried to build new partition table, misread things, used 1M EFI partition. Yeah, this is a candidate for liveboot-only. :D What I found. 1) BIG SECRET: EFI data gets saved to NVRAM, and mount reads from the NVRAM. fdisk, cfdisk, gdisk, and gparted DO NOT WRITE TO THIS NVRAM. So, when I tried up the EFI partition size, mount still loaded a 1M partition--even after reboots. How DOES it get wr | 01:37 |
AnOldHacker | ten? Clearly something in the EFI subsystem. Which leads to 2) MAJOR NON-OBVIOUS FACT: When you use fdisk, and you create an EFI partition, and it warns you about a flag, DO NOT CLEAR IT. Especially if you end up using gparted to set the partition type to VFAT32. Which gets us to 3) SECRET: If you use gparted to set the partition type, refracta will think that the drive is formatted. Grub-install will then fail with "Unknown Part | 01:37 |
AnOldHacker | ion Type" which is _really_ confusing, because gparted will _look_ _exactly_ _like_ _the_ _pictures_ _say_. Sure, there are warnings "do not format an existing efi partition", but what exactly does that mean when you yourself just created the partition? | 01:37 |
AnOldHacker | So in the end, I used FDISK to create a 550M partition, changed its type to 1, ignored the warning about the EFI flag, created a second partition for the rest of the drive, and ran refracta. The EFI install "just worked"--so long as I didn't drop into the chroot & cause that part of the process to be skipped. | 01:37 |
gnarface | AnOldHacker: if you want the saga to be preserved for posterity, you could put it on the forum and perhaps it can serve as a warning to others | 01:41 |
AnOldHacker | Pointer? | 01:41 |
gnarface | it's in the channel topic | 01:41 |
AnOldHacker | Well, if I can serve as a warning to others.... :D | 01:42 |
AnOldHacker | gnarface: Heh. I realized I forgot all about the multi-media install path issue above. If you want to pile anything on, here it is: https://dev1galaxy.org/viewtopic.php?pid=45198#p45198 | 02:00 |
rwp | AnOldHacker, I am glad to hear that you got things figured out! Thank you for coming back here and reporting how things worked out for you. | 03:50 |
AnOldHacker | I'm a firm believer in the obverse of principle of the conservation of pain--if you've felt the pain, you should be able to help others avoid it! | 03:52 |
rwp | As to a couple of details, yes, half of what is needed for UEFI to boot is the files on disk and the other half is magic pixie dust stuff stored in NVRAM. I don't understand the details either. | 03:53 |
rwp | One can use efibootmgr to view and modify the NVRAM data. That's new with UEFI boot as compared with Legacy BIOS boot which does not have anything similar. | 03:54 |
rwp | A wishlist item from me for the installer would be to have the installer create ALL of the magic partitions needed for booting any system, such that the disk could later be pulled and moved to another system hardware and then be made to boot there. But no joy. If one wants that then they need to figure out the partitions themselves. | 03:55 |
rwp | As far as I know there are at least three partition layouts required. 1) Legacy BIOS 2) UEFI 3) UEFI booting Legacy BIOS using the CSM compatibility service module. And each of those has different partition requirements. | 03:56 |
AnOldHacker | It might actually be less work to identify the bios & firmware version, and build to that. *Might*. My time at AMD 20 years ago strongly suggests that the BIOS manufacturers cannot be bothered to hire competent people to program their parts. I strongly doubt there has been a change. | 03:58 |
rwp | My HP workstations all have buggy UEFI with various different bugs even though they are otherwise closely similar. | 04:02 |
rwp | That's just another confirmation that the large corporate houses can't get it right. | 04:03 |
AnOldHacker | I would be strongly that HP does not program the BIOS in their computers. The BIOS manufacturer does that. | 04:07 |
rwp | When I worked there we definitely had a firmware team that was in charge of it. They licensed it and then tweaked it but they were the last ones to touch it. | 04:12 |
AnOldHacker | Huh. Well, that's an improvement. | 04:12 |
onefang | rwp: I have something called Magic Pixie Dust, 20+ operating systems on a micro SD card, including a variety of rescue tools. Also micro-SD to SD and USB converters / readers. | 05:31 |
onefang | I'm in the middle of rewriting my usual install script to use Daedalus, next is to redo Magic Pixie Dust for Daedalus, and adding any more things that can help this sort of thing. I'm also looking at a BIOS thing that can pretend to be EFI so I can boot with rEFInd on BIOS systems. | 05:34 |
rwp | If I were in offtopic I might mention that most stories I have read about pixie dust describe it as very itchy in the extreme for a few days. But we aren't so I won't. :-) | 06:34 |
AnOldHacker | rwp :D | 06:58 |
rrq | onefang: legacy bios chain loading boot via UEFI bios emulation sounds like a Project :) | 07:52 |
onefang | DUET looks interesting for that. https://www.rodsbooks.com/bios2uefi/ | 08:07 |
rrq | right; is tianocore in different name? | 08:19 |
rrq | hmm; moved to microsoft (github) ? | 08:20 |
rrq | .. and then leads to "Clover EFI bootloader" as the current | 08:25 |
rrq | no that's also moed to microsoft (github) | 08:25 |
onefang | I haven't actually looked deeper than that page yet. | 08:30 |
rrq | ok.. then you might warp to CloverHackyColor/CloverBootloader directly :) | 08:33 |
rrq | hmm doesn't stick in the eye as boot equipments for a hybrid ISO; rather an end-users boot loader to handle multiple installations | 08:44 |
rrq | (.. and heavily apple-fanboi by the look of it) | 08:46 |
onefang | If you find something that does the job, lemme know. I'm busy with my daedalus install script this week. | 08:48 |
onefang | Nothing wrong with apple-fanboi. Oh wait, you don't mean the fruit or the music company. Paul McCartney played near here last night. | 09:01 |
onefang | rrq: Ah Tianocore is the EFI BIOS used with qemu. | 12:01 |
onefang | ovfm package - "Open Virtual Machine Firmware is a build of EDK II for 64-bit x86 virtual machines. It includes full support for UEFI, including Secure Boot, allowing use of UEFI in place of a traditional BIOS in your VM." At boot time it shows a Tianacore logo. | 12:05 |
rrq | does it work as "bootloader" on legacy boot ? | 12:11 |
onefang | It works as qemu's EFI BIOS. Haven't tried to see if I can chain it from a legacy BIOS. | 12:16 |
rrq | as bios it resides in the boot rom while as bootloader it resides on a device; it might get confused esp about initialising its residence device | 12:17 |
onefang | Well you mentioned Tianocore and I saw the logo when booting this qemu in EFI mode. So it's a BIOS, not a bootloader in this case. | 12:20 |
rrq | afaiui tianocore is the base for most current uefi implementations, and it's also what's used in "clover"; so it'd be good as uefi emulator if it'd be happy residing on a device | 12:22 |
rrq | or else it needs a prior loader that "installs" it fully into ram | 12:25 |
rrq | and then that loader needs device reading ability, handling all sorts of devices | 12:28 |
rrq | egg meets chicken | 12:29 |
onefang | So I'll look at this clover thing later. | 12:29 |
strato | I have problems with wine in daedalus. Is this a known issue, or do I have to dig deeper? | 13:39 |
djph | depends on "the problem" (I mean, I'm not having any problems with the few wine-games I have) | 13:41 |
strato | This is one of the errors I get: | 13:41 |
strato | 00d4:err:module:load_wow64_ntdll failed to load L"\\??\\C:\\windows\\syswow64\\ntdll.dll" error c0000135 | 13:41 |
djph | so WoW is missing a dll, or it's broken | 13:42 |
djph | https://forum.winehq.org/viewtopic.php?t=37610 <- first hit on google | 13:43 |
djph | didn't even include "wine" in the thing | 13:43 |
strato | What is WoW ? | 13:43 |
djph | I thought it was "World of Warcraft" but ... um, maybe not :) | 13:44 |
strato | My google hit was, among others: https://forum.winehq.org/viewtopic.php?t=36023 :) | 13:45 |
strato | I tried it, it dont work. | 13:45 |
djph | :( | 13:45 |
djph | maybe i'm just lucky because I'm running a chimaera -> daedalus upgrade ? :| | 13:46 |
strato | It did run before I upgraded | 13:46 |
strato | Well, it does not seem to be a known issue, so I dig deeper :) | 13:48 |
KZSAZO | djph: https://bugs.launchpad.net/ubuntu/+source/wine/+bug/2024267 you're missing the 32bit wine packages | 14:48 |
KZSAZO | make sure to check out the WineHQ repo too https://wiki.winehq.org/Debian | 14:49 |
djph | KZSAZO: I'm not missing anything, it's working here. but maybe that'll help strato | 14:54 |
KZSAZO | djph: oh, i meant to quote strato :) sorry! | 14:56 |
djph | :) | 14:57 |
djph | KZSAZO: the keys are like right next to each other | 14:57 |
strato | I did install and uninstall wine32 several times, it did not help, so I am looking further. | 15:28 |
strato | And by the way: lscpu | 15:34 |
strato | Arkitektur: x86_64 | 15:34 |
strato | Op-tilstande for CPU: 32-bit, 64-bit | 15:34 |
strato | So 32 bit architecture should be activated | 15:35 |
gnarface | strato: show me the output of these commands (put them at paste.debian.net): | 16:05 |
gnarface | dpkg --print-foreign-architectures | 16:05 |
gnarface | dpkg --print-architecture | 16:06 |
gnarface | dpkg -l |grep wine | 16:06 |
gnarface | just a sanity check first | 16:06 |
gnarface | btw the ntdll thing is probably a red herring | 16:06 |
gnarface | djph: completely confusingly, "WoW" is also the acronym microsoft uses to describe their dual-architecture dll environment | 16:08 |
gnarface | it's their equivalent of "Multiarch" though it's a complete mess by comparison | 16:09 |
gnarface | (basically every piece of software indiscriminantly uses a mix of both architectures) | 16:10 |
djph | gnarface: I am ever more glad I'm not using windows | 16:14 |
gnarface | (and it's extra more confusing when searching wine support forums about it, because microsoft's stupid "syswow64" and World of Warcraft are usually both being debugged at the same time) | 16:14 |
gnarface | World of Warcraft being one of the first major games to use this | 16:15 |
gnarface | did microsoft name it that way to purposefully cause confusion? there's no other plausible explanation | 16:15 |
gnarface | strato: btw there's some package in the repos you can install to get rid of the ntdll error, i'm trying to remember the name... but like i said, usually that's not the actual problem. | 16:16 |
gnarface | that error often shows up even when stuff all works right | 16:16 |
gnarface | damnit | 17:20 |
gnarface | i just remembered the package name is winbind, but strato is already gone | 17:20 |
gnarface | if strato comes back, tell him the package name i was trying to remember above is winbind | 17:21 |
gnarface | but like i said, that'll probably suppress the error but not the actual problem | 17:21 |
aitor_ | hi | 20:42 |
aitor_ | today i've gotten thunar-volman's volume management working with vdev for the first time, and the *.pkla files defined for udisks2 in /etc/polkit-1/rules.d are working as well | 20:46 |
aitor_ | thanks to the recent change in the helpers: | 20:50 |
aitor_ | https://github.com/storaged-project/udisks/discussions/1209 | 20:50 |
plasma41 | aitor_: Neat! :-) | 20:50 |
aitor_ | now i'm planning to unify both libeudev and libudev-compat into one shared library | 20:51 |
plasma41 | aitor_: Sounds great | 20:51 |
aitor_ | I think it's possible, and advisable, because change | 20:51 |
aitor_ | ... because changes between them, while possible, are a bit of a complication due to libapt's dependency on libudev | 20:52 |
plasma41 | I'll take your word for it. The {,e}udev code goes a bit over my head, but it's reassuring to hear that it's being worked on by someone knowledgeable about it. | 20:55 |
aitor_ | eudev is being mantained by Boian Bonev, among other contributors | 20:57 |
aitor_ | actually eudev is working better than vdev, but I find the latter very promissing | 20:59 |
plasma41 | aitor_: I like having options, so I encourage the experimentation with multiple projects | 21:02 |
aitor_ | yes, this is what i'm trying to do, I also started including s6 in gnuinos | 21:04 |
aitor_ | and I like it very much | 21:05 |
plasma41 | aitor_: Do you know what the state of s6 is in Debian? https://tracker.debian.org/pkg/s6 I've not had the opportunity to try it out. | 21:06 |
aitor_ | a bit stuck | 21:07 |
aitor_ | some packages are missing | 21:07 |
aitor_ | I packaged the most recent versions for gnuinos, but they have not been uploaded to the repository yet | 21:08 |
aitor_ | however, there is an iso image for testing | 21:08 |
plasma41 | aitor_: Link? | 21:08 |
aitor_ | https://www.gnuinos.org/mirror/daedalus/s6%20(experimental)/ | 21:09 |
plasma41 | Thanks :-) | 21:09 |
aitor_ | not for production, of course | 21:09 |
plasma41 | Understood | 21:09 |
aitor_ | all the stuff related to sysvinit has been removed, it's not an hybrid init | 21:10 |
aitor_ | pure s6 | 21:10 |
plasma41 | Gotcha | 21:10 |
aitor_ | you'll not find things like /etc/inittab and the like | 21:11 |
aitor_ | plasma41, time to dinner | 21:11 |
aitor_ | i'll be in touch | 21:11 |
aitor_ | here, more often... | 21:12 |
plasma41 | Thanks for chatting with me. TTYL. Enjoy your meal. | 21:12 |
aitor_ | thanks, bye devuaners :) | 21:12 |
plasma41 | \o | 21:13 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!