channel logs for 2004 - 2010 are archived at http://tunes.org/~nef/logs/old/ ·· can't be searched
#osdev2 = #osdev @ Libera from 23may2021 to present
#osdev @ OPN/FreeNode from 3apr2001 to 23may2021
all other channels are on OPN/FreeNode from 2004 to present
http://bespin.org/~qz/search/?view=1&c=osdev&y=19&m=8&d=16
03:25:22 <XainFaith> Hopfully someone with experince with the pl181 is on and could tell me why i am getting a value of 0x120 from a cmd 55
03:26:24 <XainFaith> according to the R1 response type that would be a bad address flag as well as a bit set past the bounds of the actual response size, i am using qemu-system-arm
03:26:34 <XainFaith> intergrator-cp
03:28:17 <ybyourmom> Did you execute order 66
03:29:09 <XainFaith> im trying to initalize the card so iv sent cmd 0 and cmd 8 and varified it can run off the voltage and the check pattern is good
03:29:28 <XainFaith> thats all iv done up to the cmd55
03:35:53 <XainFaith> and the status registers also looks normal
03:40:55 <ybyourmom> https://knowyourmeme.com/memes/execute-order-66
03:41:37 <XainFaith> so no i have not lol
03:42:45 <Mutabah> Remind me, what sort of chip is the PL181?
03:42:58 <XainFaith> multi media controller interface
03:43:16 <Mutabah> Ah, not worked with it sorry :/
03:43:29 <XainFaith> all good :)
03:44:50 <klys> now that I have the monitor from the mail, it's 14" not 15" so it isn't 96dpi; also there is a little burn-in on the screen from "Sears Gift Registry."
04:12:07 <while> hi, I'm trying to learn the "language" used in sixels as seen in vt340-compatible terminal-emulators what is a good reference for that?
04:12:27 <while> like is there an RFC?
04:16:48 <klys> https://vt100.net/docs/vt3xx-gp/about.htm and ftp://ftp.fu-berlin.de/unix/www/lynx/pub/shuford/terminal/all_about_sixels.txt appear to cover the sixel topic from a lay perspective.
04:35:15 <klys> so it seems the optimal resolution for this 14" display would be 1075x807
04:47:55 <while> thank you klys
04:49:45 <XainFaith> don't suppose anyone has popped on that has used the pl181
04:51:12 <zid`> I could google it for you
04:51:15 <zid`> would that help
04:51:26 <zid`> PL181 High Temperature Dry Film Lubricant
04:52:49 <klange> i'm getting a tv stand
04:52:55 <zid`> qemu doesn't really do much
04:53:07 <zid`> it does one sanity check then returns a value
04:53:26 <klange> but I *assume* XainFaith means this arm 'multimedia card interface'
04:53:32 <XainFaith> arm prime cell pl181 mmci
04:53:36 <XainFaith> specifically
04:54:05 <XainFaith> well still the value its giving back makes no sense in regards to what it should be thats the problem really
04:54:32 <zid`> well it returns either r1
04:54:51 <zid`> also it's case 55 not case 0x55 just to double check?
04:55:01 <zid`> https://github.com/hackndev/qemu/blob/master/hw/sd.c#L982https://github.com/hackndev/qemu/blob/master/hw/sd.c#L982
04:55:01 <XainFaith> the R1 Response that it returns is one bit larger then it should be
04:55:02 <zid`> oops
04:55:04 <zid`> https://github.com/hackndev/qemu/blob/master/hw/sd.c#L982
04:55:13 <XainFaith> and yeah im not using hex values for the cmd value
04:55:33 <XainFaith> and cmd 55 R1 reponse says its a bad address error flag
04:55:48 <XainFaith> for the bit that does fit within the size of the R1 response
04:56:41 <XainFaith> yeah i see that i wounder if why then i get the value i do
04:57:00 <zid`> fime for a teabugger
04:57:05 <XainFaith> the RCA by default should start off at all zeros
04:57:14 <zid`> else it gets set to += 0x4567
04:57:17 <zid`> in some other case
04:57:22 <zid`> appears to be the only place it changes
04:58:26 <XainFaith> hmm
05:01:56 <XainFaith> yeah it defintly aint doing a heck of a lot so i guess im going to really need a seperate driver for qemu perhaps
05:03:52 <XainFaith> eqaully though even if i push pass that the rca i get is not right ether it seems
05:04:02 <zid`> what is the rca anyway
05:04:03 <XainFaith> so something some where is not quite right
05:04:39 <XainFaith> well i send cmd55 followed by 41 and i get back 0xFFFF00
05:08:00 <XainFaith> .. i think i might have an idea of what it is
05:09:33 <while> what advantage did the antenna port on "modern" analog televisions have over the 70s "twin-turbo" cabling?
05:09:58 <while> the coaxial vs balanced pair
05:10:11 <Mutabah> ... not exactly on-topic, but I suspect it gave simpler integration with coaxial cables
05:10:21 <Mutabah> and coax is much better at noise resiliance
05:10:52 <while> ik, but it seems like all the useful people on freenode congregate here
05:11:10 <XainFaith> i figured maybe i was missing a clear of flags but thats not it
05:11:26 <XainFaith> well ill have to just really pour over that the qemu code for the pl181 and see if i can't come up with whats going on
05:11:36 <XainFaith> but thanks at least with that i might get some where
05:11:38 <zid`> debuggers are things
05:11:42 <zid`> don't forget
05:11:44 <XainFaith> i ave gdb
05:11:55 <XainFaith> but that does not help me figure out why qemu is giving me the value it does
05:11:56 <zid`> hope you built qemu splitdebug
05:12:01 <XainFaith> nope but i can
05:12:04 <while> zid` rca is specifically a connector type, common place on component video, L-R analog audio, and composite video
05:12:11 <while> cables
05:12:16 <zid`> while: absolutely nothing to do with the topic
05:12:24 <XainFaith> LOL
05:12:26 <while> sorry, I saw you ask earlier
05:12:44 <zid`> also sad that you think I'd be that dumb
05:12:58 <XainFaith> rca in this case has to do with sd card relative address
05:13:02 <XainFaith> so to speak
05:13:22 <while> zid` the new generation of people growing up doesn't even recognize public phone booths!
05:13:54 <while> and I had to explain to my brothers what "that square where the save button should be is"
05:14:19 <while> knowledge is disappearing before our eyes quickly...
05:14:41 <bluezinc> while: BS.
05:14:57 <zid`> when google stops giving the right answer you may complain
05:15:10 <zid`> until then, young people are just young and don't know much
05:15:16 <bluezinc> plenty of people under 20 know about phone booths, floppy disks, etc.
05:16:05 <while> most between 15 and 20 do, but it is a growing 'problem'
05:16:38 <zid`> it is not a problem
05:16:46 <bluezinc> knowing what a floppy disk is doesn't make you old. thinking that nobody else knows does.
05:17:09 <bluezinc> arguably, it's about as much of a problem as you not being able to recognize a barber's leech jar from centuries past.
05:17:18 <zid`> I don't know how to fix a jacquard loom, boo hoo
05:17:52 <bluezinc> zid`: 'tis a shame, the modern woman can no longer fix a spinning wheel. :P
05:18:00 <zid`> disgraceful
05:18:07 <while> yeah...
05:18:35 <bluezinc> how are they going to prick their finger (I think there's a witch involved!) and fall asleep for a hundred years!
05:18:55 <zid`> then get raped by every passing prince
05:19:29 <bluezinc> * the king formerly known as prince
05:21:38 <bluezinc> zid`: I think my favorite case of this was some comment by a schoolmaster to the effect of "kids these days with their paper! back in my day we used slates!"
05:21:48 <zid`> yea I actually saw that yesterday
05:21:55 <zid`> it gets reposted a lot when people moan about children or tech
05:22:21 <bluezinc> if we had better surviving sumerian records, I'd assume we'd hear people moaning about the wheel.
06:05:19 <XainFaith> turns out this may be a bug with the build of qemu i have right at the moment
06:05:49 <zid`> git blame
06:06:31 <XainFaith> well some others have already posted bug reports and i have yet to debug it myself but that defintly looks like what i am going to have to do
06:07:12 <XainFaith> q
16:12:57 * geist yawns
16:13:09 <geist> good morning folks
16:14:34 <Prf_Jakob> Morning _o/
16:19:19 <zid`> geist: I couldn't find anything to throw into your gaping maw in time, sorry
16:34:20 <Ameisen> still find it odd that I had to report the same performance bug for llvm, gcc, and msvc
16:34:44 <Ameisen> where __assume or equivalent breaks tail call optimizations
16:36:55 <_mjg> burp
16:37:06 <_mjg> you guys look like crap, you okay?
16:39:58 <geist> Ameisen: didn't you do that yesterday?
16:48:49 <Ameisen> geist - yes, but I had to respond to some comments today
16:48:53 <Ameisen> so it's still fresh in my mind
16:51:27 <geist> nice, good to see that at least some of them are interested
16:51:47 <geist> re: clang, there was some internal emails about a new pass manager in clang that isn't yet default but will become so
16:52:06 <geist> and iirc one of the bullet point items of it is it'll start paying attention to likely/unlikely
16:52:15 <geist> which i've personally observed clang doesn't really follow
16:52:21 <geist> may be similar for your stuff
16:55:51 <Ameisen> well, the MSVC comment was basically "how do we reproduce this"
16:56:04 <Ameisen> to which my response was basically "put it in a source file, compile, and look at the assembly."
16:56:39 <Ameisen> Well, __builtin_assume (and such) aren't branch prediction hints but are rather value assertions
16:56:42 <Ameisen> which is why it breaks
16:57:04 <Ameisen> if you do int a = pure_function(); __builtin_assume(a == 1); if (a == 1) function();
16:57:21 <Ameisen> their IR replaces their internal 'a == pure_function()` with 'a == 1'
16:57:35 <Ameisen> err, return a; as well
16:57:47 <Ameisen> which has a cascading effect of causing it to not realize it can in the end be optimized to a tail call or such
16:57:58 <Ameisen> since it stops recognizing the variable as having been the return value of a call
16:58:09 <geist> ah
16:58:16 <Ameisen> it's just sorta funny that an optimization hint breaks things
16:58:45 <Ameisen> I wouldn't be surprised if it was the same reason on all three, since it's sorta a tricky situation
16:59:31 <eryjus> ameisen, I thnk they all use each others' compilers as reference solutions, so if they all behave the same, then they think they have it right
17:00:01 <Ameisen> Well, it's not bad codegen per se. The code is perfectly fine. Just suboptimal.
17:00:59 <j`ey> also sounded like a bit of a corner case
17:01:12 <Ameisen> It is, but note that these are also reduced samples.
20:04:14 <zid`> klange: new best girl is yamamoto from joshikousei no mudazukai, is she going on your car
21:02:58 <Ameisen> reported another one. I think I just write really weird code.
21:03:52 <griddle> There needs to be a tool that you can point at a directory, and it creates a disk image without using linux's mount syscall
21:03:59 <griddle> anyone know of one?
21:05:00 <griddle> I can imagine porting a filesystem from my kernel and getting it to work with an mmapped file by emulating a block device
21:07:47 <griddle> I may do that. I kinda hate that I have to use sudo to make a rootfs
21:14:55 <froggey> mke2fs + debugfs with a little scripting can do that for ext{2,3,4}. mkisofs/genisoimage for iso9660 file systems
21:15:28 <froggey> genext2fs exists too, but only generates really old style ext2 file systems
21:20:08 <kirenoto> It uses the mount command and syscall, you can use Linux's loopback storage mechanism to have any file act as a filesystem, even as a regular user.
21:20:15 <griddle> Ext2 is all I need for now, can I point it at a folder and it copied the structure into the disk image?
21:20:33 <griddle> I want to avoid having sudo in my makefile
21:20:40 <kirenoto> As I recall, you need to grant the user access to the loop device file in /etc/fstab though.
21:21:17 <kirenoto> Likely also needs write access to /dev/loopX as well.
21:24:02 <froggey> griddle: yep, genext2fs will do exactly that. no special permissions needed at all
21:25:01 <kirenoto> Oh, note that with the loop device method you'd be setting the "user" option in fstab, and the mount point will be static/fixed in place. Trying to run mount with special non-default options or a different mount point doesn't work.
21:26:00 <griddle> froggey: thanks a bunch
21:26:44 <kirenoto> There's actually even more ways to do this than I realized ... Linux is pretty crazy in its versatility / feature set. https://unix.stackexchange.com/questions/96625/how-to-allow-non-superusers-to-mount-any-filesystem
21:27:49 <kirenoto> TL;DR is having a management daemon like GVFS makes it possible, and there's also user namespaces. My method was the lower ranked reply ... lol
21:28:08 <kirenoto> I bet there's a way to make it work with FUSE (userspace filesystems) as well.
21:29:50 <mrvn> kirenoto: users can't mount or use losetup
21:31:29 <mrvn> griddle: look into e2tools or fuse2fs
21:31:40 <mrvn> or fuseext2
21:32:19 <kirenoto> Users can mount specific devices, if allowed by fstab with the "user" option.
21:32:38 <kirenoto> They can't use losetup, but that's a one-time setup.
21:32:47 <mrvn> kirenoto: true. But that's really broken for building sources.
21:34:51 <mrvn> e2tools can manipulate the FS directly without mounting.
21:36:17 <kirenoto> Yeah, if you're not actually using it like a normal filesystem, that may be better.
23:45:06 <klange> zid`: one-off anime that won't be getting sequels are but passing fads; vocaloid is eternal
23:45:17 <klange> unrelated, go-to definition in my editor: https://asciinema.org/a/262772