Search logs:

channel logs for 2004 - 2010 are archived at ·· 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

Friday, 14 January 2022

03:31:00 * kazinsal cheers
03:32:00 <kazinsal> bright red angry TEST FAILURE on this server diagnostic means I get to throw this back at Cisco TAC and go "I told you the CPU was fucked"
03:36:00 <blockhead> non-working CPU is reason to cheer?
03:36:00 <Mutabah> it's somebody else's problem
03:36:00 <kazinsal> it is when you've been fighting with the vendor for months
03:36:00 <blockhead> aha! Understood!
03:37:00 <Affliction> Well, dealing with cisco TAC, good luck having them take anything less
03:37:00 <kazinsal> yeah, I work for a cisco gold partner and it's amazing how much shit they put us through to get stuff that isn't like, a phone replaced
03:38:00 <kazinsal> like, come on guys, we sell eight figures of your hardware per year. give us a break on RMAing a friggin BE6K
03:39:00 <heat> me: ports rust
03:39:00 <heat> rust the day after: hey guys new version
03:39:00 <heat> fml
04:04:00 <klange> I think gcc 11 released right around when I got 10.3 all set up.
04:06:00 <klange> How I know I'm really unimportant... no one has bothered to add toaru to autotools.
04:07:00 <klange> Wikipedia rules. If I do it, it's pretentious and self-aggrandizing, so someone else has to.
04:08:00 <heat> yeah same
04:08:00 <heat> sortie told me to add it but I dont feel comfortable adding my own OS to a list of OSes
04:09:00 <heat> i wish upstream would take my OS's patches
04:10:00 <heat> i'd never need to maintain a patch again
04:12:00 <klange> Especially with all the stuff using autoconf where the only change I ever need to make is adding another switch case...
04:16:00 <heat> those are annoying but not the worst IMO
04:16:00 <heat> the worst is stuff like LLVM where you need to add your OS everywhere (code, build system, whatever) and things break slightly every other release so you can't reuse the same patch
04:16:00 <heat> .rej files everywhere
04:17:00 <klange> Yeah, gcc patch updates are not fun which is why I was on, eg, 4.6 for a very long time, then 6.4...
04:18:00 <klange> And 10.3 only happened for the new kernel since I had to configure for a new architecture anyway...
04:18:00 <heat> and then the bulk of my LLVM patch (sanitizer stuff) is kind of applicable to GCC's libsanitizer and kind of not because GCC doesn't really use the latest LLVM release's sanitizer code
04:18:00 <heat> and they apply some really weird patches to it, which I can't find
04:19:00 <heat> so I end up kind of needing to maintain two huge patches of kind of the same code
04:19:00 <heat> brilliant, 10/10
05:40:00 <klys> if anyone here is using xen, I might like to talk it over with you.
05:41:00 <klys> I have a ridiculous qubes setup that works to a point
05:41:00 <klys> and am planning on using it perhaps without all the python
05:44:00 <klys> also, running qemu with xen from a practical standpoint might interest someone less familiar
07:31:00 <kazinsal> xen is one of those things I still haven't actually played with
07:31:00 <kazinsal> and I don't really have any hardware to install it on unfortunately
07:43:00 <klys> also if someone is barely interested, perhaps a good first step would be putting xen in a vm testbed. this page shows to run qemu with a couple of options.
07:43:00 <bslsk05> ​ Nested Virtualization in Xen - Xen
07:43:00 <klys> I've been looking for the options actually
13:20:00 <gh555222> are there any circumstances in which using a cross compiler would be bad? I'm new to the principle of cross compiler and was mostly learning from here:
13:21:00 <bslsk05> ​ Why do I need a Cross Compiler? - OSDev Wiki
13:22:00 <sham1> I can't think of any negatives from using a cross-compiler
13:23:00 <GeDaMo> As long as it generates correct code for the target there shouldn't be any problems
13:24:00 <GeDaMo> Machine code is machine code :P
13:24:00 <sham1> Not really
13:27:00 <gh555222> GeDaMo: from what I read it looks like the biggest problems come out as the project gets bigger
13:28:00 <gh555222> mostly the need for platform-dependent compilation switches that may need to be different between platforms (a cross-compiler would I guess solve this by being the "one true" compiler)
13:29:00 <sham1> Yeah, you can customise the cross-compiler in ways that you mightn't be able to with a non-cross-comper
13:30:00 <gog> the thing that happens when you use the host compiler is that it might pull symbols from host libraries. -ffreestanding and -fnostdlib prevent that mostly
13:31:00 <gog> but the best guarantee against any code that depends on host libraries is a cross compiler
13:31:00 <zid> -nostdlib, no f
13:31:00 <gog> yes
13:32:00 <sham1> Well, a generic cross compiler. One could still build a cross compiler for Linux and still be wrong
13:32:00 <zid> a cross compiler is no different to any other compiler, anyway, it's just whose machine ends up running it
13:32:00 <gog> yes
13:33:00 <zid> Simpler example: nasm is nasm, regardless of whether you compile it for linux or windows, it still functions the same and outputs the same shit.
13:45:00 <gh555222> isn't it still somewhat likely 3rd parties with limited quality assurance would write assembly code depending on the host platform in the absence of a cross compiler/assembler?
13:46:00 <zid> I have no idea what *any* of that sentence means
13:46:00 <zid> structurally, semantically, syntactically
13:46:00 <gh555222> if a cross compiler prevents issues related to host-platform dependency
13:47:00 <gh555222> wouldn't using a portable compiler pose similar issues to not using a cross compiler?
13:47:00 <zid> define 'portable compiler'
13:47:00 <gh555222> you mentioned 'nasm'
13:47:00 <zid> am I having a stroke
13:47:00 <gh555222> okkkk, I think i've overstayed my welcome
13:48:00 <zid> no I just genuinely
13:48:00 <zid> okay
13:49:00 <gog> :|
13:50:00 <gog> this is osdev, who does QA?
13:50:00 <sham1> You could have just said that nasm is a cross compiler. It's not totally true, but it's correct-ish
13:50:00 <zid> why, in response to *what* lol
13:50:00 <zid> I genuinely have 0 clue what any of his sentences meant
13:51:00 <sham1> To the portable compiler question.
13:51:00 <kingoffrance> portable: someone else already wrote the assembly, or interfaced with the platform-dependent parts, to provide an illusion
13:52:00 <kingoffrance> i find that type of definition works best :)
13:52:00 <gog> i write low-quality assembly maybe they meant me
13:52:00 <sham1> I think what they meant was to ask "why doesn't one need a nasm cross compiler"
13:52:00 <zid> I don't understand that either
13:52:00 <klange> I consider being portable more of a measure of how possible it is to port something
13:52:00 <zid> same klange
13:52:00 <sham1> zid: well then you're lost
13:52:00 <zid> I knew that already sham
13:53:00 <klange> So, a compiler that emits assembly directly and is very tied to the format is not portable because porting it requires rewriting essentially all of it
13:53:00 <kingoffrance> i got very portable code. compiles everywhere. 1000 dummy /no-op backends yet to be implemented. 100% portable, 100% nonfunctional
13:53:00 <zid> also in assemblers/compilers you have portable in the sense of retargetable, on top
13:53:00 <zid> two axes of portable
13:54:00 <gog> euclidian portability
13:55:00 <zid> I thought that was cartesian
13:55:00 <zid> I am not a math though
13:56:00 <gog> oh
13:56:00 <gog> it is
13:56:00 <gog> euclidian is 3d
13:57:00 <sham1> R^n in general
13:57:00 <zid> is cartesian cognate to cartography?
13:57:00 <gog> no it's cognte to René Descartes
13:57:00 <sham1> The D!
13:57:00 <zid> right, but does the des cartes mean "the maps" :P
13:58:00 <gog> maybe it was a nickname he gave himself because he love map
13:58:00 <GeDaMo> # Old Descartes was a drunken fart, 'I drink therefore I am' # :P
13:58:00 <zid> could be
13:58:00 <zid> gog thebutter
13:58:00 <kingoffrance> i thought atlas was in charge of atlases
13:59:00 <gog> i do love butter. on a bagel
13:59:00 <sham1> Butter tends to do that
14:06:00 <zid> Got me a kilo of finest ungogged butter right from the port
14:32:00 <sham1> s/ungogged/ungoogled/
18:03:00 * geist yawns
18:03:00 <zid> I accidentally the hot
18:04:00 <zid> My medcine bottle of chilli extract got a little bit excited when I added some
18:11:00 * gog slides a coffee to geist
18:12:00 <geist> just a little. going into the doc for a yearly checkup so been fasting since last night
18:12:00 <geist> but a little bit of black coffee is okay
18:12:00 <geist> but gonna be so hungry in a little bit
18:13:00 <gog> heh when i went the other day i had to run to the clinc from the bus stop to not miss my time. my bp was a little elevated
18:15:00 <GeDaMo> My aunt is a nervous personality, she has to take her own bllod pressure at home because just going into the doctor's waiting room sends it through the roof :P
18:15:00 <geist> yah there's some effect of doc offices where BP is generally elevated due to doc anxiety
18:15:00 <geist> yah
18:15:00 <gog> yeah i think i have that too
18:16:00 <zid> I don't care about doctors, receptionists are the problem
18:16:00 <gog> ok g2g see a gal about a dog
20:41:00 <heat> hell
20:41:00 <heat> o
22:37:00 <heat> if I already take a hit acquiring a lock (with atomics) is doing an atomic operation while holding a lock as expensive as it usually is?
22:37:00 <heat> example: lock a spinlock, atomically increment a refcount, unlock