#archlinux32 | Logs for 2023-04-18
Back
[00:31:11] <ConSiGno> welp, my HDD died! reinstalling on a CF card. had to pass "libata.dma=3" to the kernel.
[06:41:55] -!- GNUtoo has quit [Remote host closed the connection]
[06:42:12] -!- GNUtoo has joined #archlinux32
[12:14:51] -!- abouvier has quit [Quit: kthxbye]
[12:15:15] -!- abouvier has joined #archlinux32
[13:08:20] -!- zxrom has joined #archlinux32
[13:32:45] -!- T`aZ has quit [Ping timeout: 240 seconds]
[13:34:52] -!- T`aZ has joined #archlinux32
[14:29:05] <ConSiGno> back! Turns out that turning off DMA for CF cards is the right thing to do. no more errors :D
[14:35:44] <KitsuWhooa> I think it depends on the CF card
[14:36:39] <ConSiGno> it's getting to the point that cards are "too fast"
[14:36:45] <ConSiGno> so you have to slow them down
[14:37:40] <ConSiGno> but now I have a super quiet PC so that's good :)
[14:37:41] <KitsuWhooa> linux isn't using the BIOS routines to access the CF cards though, right?
[14:38:20] <KitsuWhooa> same with disk size limitations IIRC
[14:38:40] <KitsuWhooa> as long as the kernel boots (say, boot partition is within the bios limit), it then works fine
[14:56:38] <ConSiGno> no it doesn't use the bios if it doesn't have to
[14:57:15] <ConSiGno> you can write programs that do that but that's what the ABI is for
[14:57:37] <ConSiGno> if you write for bios you make your program incompatible in the future
[14:58:44] <KitsuWhooa> not sure about that :p
[14:58:44] <ConSiGno> it's one of the reasons PC Clones were really hit or miss in the old days. if you wanted full speed you wrote directly for the BIOS, but you could only run your software reliably on an IBM PC. If you're looking for old PC clones it's something to keep in mind
[14:59:03] <ConSiGno> well maybe you can't do that anymore (I hope not!)
[14:59:13] <KitsuWhooa> DOS uses BIOS for I/O
[14:59:26] <KitsuWhooa> (I think MS-DOS specifically)
[14:59:39] <KitsuWhooa> and you can have real trouble with CF cards on some broken BIOSes
[15:00:17] <KitsuWhooa> so yeah, I'm not sure why the "CF card is too fast" to cause issues in your case, since linux won't be using that
[15:00:46] <KitsuWhooa> if it does DMA, isn't it going to be UDMA33 anyway?
[15:01:19] <ConSiGno> I think this one is 66
[15:02:21] <KitsuWhooa> and you're using an 80 conductor cable (assuming the card isn't plugged in directly to the port)? :p
[15:02:39] <ConSiGno> yeah I'm not directly plugged in :V
[15:02:44] <KitsuWhooa> Although I didn't know UDMA66 was a thing on i486
[15:02:57] <ConSiGno> this is not really a 486, this is a Socket 7
[15:03:07] <KitsuWhooa> ah
[15:03:27] <ConSiGno> 512KB L2 cache
[15:03:48] <KitsuWhooa> either way, if UDMA66 is unreliable, it sounds like a cable issue
[15:04:09] <ConSiGno> it _might_ be a cable issue, these cables are pretty old
[15:04:16] <KitsuWhooa> is it an 80 conductor cable though? :p
[15:04:45] <ConSiGno> yup
[15:05:05] <KitsuWhooa> maybe the card just can't handle cables in general I guess
[15:05:12] <KitsuWhooa> they're not really meant to :p
[15:05:47] <KitsuWhooa> or I guess the card just has broken dma
[15:06:08] <KitsuWhooa> I would first try forcing UDMA33 though
[15:06:23] <KitsuWhooa> going straight to PIO seems rough
[15:08:20] <ConSiGno> it's not as slow as I thought it would be
[15:08:37] <KitsuWhooa> it's not just about speed though
[15:08:39] <KitsuWhooa> you're stealing cpu cycles
[15:08:47] <KitsuWhooa> er, not stealing
[15:08:49] <KitsuWhooa> but you know, wasting
[15:30:35] -!- abaumann has joined #archlinux32
[15:30:35] <buildmaster> Hi abaumann!
[15:30:35] <buildmaster> !rq abaumann
[15:30:36] <phrik> buildmaster: <abaumann> wait... go-go-gadgeto-i686-test-machine
[15:55:43] -!- pulec has quit [Read error: Connection reset by peer]
[15:59:02] -!- pulec has joined #archlinux32
[16:16:32] -!- abaumann has quit [Quit: leaving]
[18:38:35] -!- T`aZ has quit [Ping timeout: 264 seconds]
[18:39:46] -!- T`aZ has joined #archlinux32
[19:23:05] <ConSiGno> hopefully I can figure out the exact DMA settings for this CF card and apply them with a boot param
[19:40:32] <ConSiGno> ... or I can get this https://www.startech.com
[19:40:33] <phrik> Title: 40-Pin IDE to SATA Adapter Converter - Drive Adapters and Drive Converters | StarTech.com (at www.startech.com)
[19:42:01] <KitsuWhooa> funny
[19:42:12] <KitsuWhooa> I have a crappy adapter on my P3 router to have a SATA SSD on it
[19:42:19] <KitsuWhooa> It isn't that reliable
[19:43:00] <KitsuWhooa> Also, keep in mind, if you plug in large drives they probably won't work on a machine of that era
[19:43:17] <KitsuWhooa> smallest SSDs you can typically get these days are 120GB
[19:47:01] -!- T`aZ has quit [Ping timeout: 240 seconds]
[19:48:50] <KitsuWhooa> 22:42:01 <KitsuWhooa> funny <-- funny as in, I was thinking of replacing it with a CF card
[19:48:50] -!- T`aZ has joined #archlinux32
[19:48:56] <KitsuWhooa> got distracted :p
[19:53:53] <KitsuWhooa> That said, I might have to take what I said back. Looking at dmesg I don't see any IO errors
[19:53:59] <KitsuWhooa> maybe it's because it rarely touches the disk
[19:59:25] <ConSiGno> I've since learned from some retro dudes on discord that Samsung doesn't really give a damn about IDE support on newer CF cards, and any errors I'm finding is due to that
[19:59:57] <KitsuWhooa> maybe?
[20:00:01] <ConSiGno> soooo I'm going to get a SATA to IDE adapter with a cache on it and firmware that I can update and be _done_ with it :V
[20:07:27] <ConSiGno> thankfully the mobo I have is pretty popular so hacked BIOSes are not uncommon
[20:08:00] -!- JerryXiao has quit [Ping timeout: 260 seconds]
[21:30:05] -!- T`aZ has quit [Ping timeout: 240 seconds]
[21:30:47] -!- JerryXiao has joined #archlinux32
[21:32:05] -!- T`aZ has joined #archlinux32
[22:03:09] <ConSiGno> really need xf86-video-mach64 back some day, I'm gonna compile it manually D:
[23:22:07] -!- T`aZ has quit [Ping timeout: 260 seconds]
[23:23:46] -!- T`aZ has joined #archlinux32