#asahi-dev · 2023-03-12 20:03
axboeand#asahi-dev · 2023-03-12 20:03
axboe[ 2.130048] /soc/pcie@590000000/pci@2,0 interrupt-map failed, using interrupt-cont#asahi-dev · 2023-03-12 20:03
axboefrom...#asahi-dev · 2023-03-12 20:03
axboeget a few of these on the current kernel, known?#asahi-dev · 2023-03-12 20:03
axboe[ 2.203371] WARNING: CPU: 3 PID: 66 at drivers/of/irq.c:279 of_irq_parse_raw+0x504#asahi-dev · 2023-02-15 00:18
axboeexcellent#asahi-dev · 2023-02-15 00:17
axboemarcan: yep, just CC me on all of them from the get-go#asahi-dev · 2023-02-14 23:56
axboemarcan: cc me on these things, I'm quite serious when I say I think this is a big deal and if my voice can add a bit of push back to the BS, then I'm more than happy to do so. I absolutely hate for this to be an unnecessary burden, seems like such a waste of time when that energy could be spent so much better on solving actual issues#asahi-dev · 2023-01-25 15:32
axboemaybe rdisk would be better, I know very little about osx#asahi-dev · 2023-01-25 15:31
axboeso suspect both are essentially buffered, just !F_NONCACHE not persistently so#asahi-dev · 2023-01-25 15:31
axboebut googled F_NONCACHE, and it's not zero copy, it just reduces caching effects#asahi-dev · 2023-01-25 15:30
axboe./fio --name=test --time_based --ramp_time=5s --timeout=30s --bs=256k --rw=read --direct=1 --size=4g#asahi-dev · 2023-01-25 15:30
axboethis is what I ran:#asahi-dev · 2023-01-25 15:30
axboeof course I used fio :)#asahi-dev · 2023-01-25 15:12
axboeoh well, will test once linux boots on it#asahi-dev · 2023-01-25 15:11
axboe15.8GB/sec F_NONCACHE, 16.3GB/sec normal buffered#asahi-dev · 2023-01-25 15:10
axboegetting ~16GB/sec on the m2 max#asahi-dev · 2023-01-25 15:10
axboenow I'm pondering if the cache bypass works on osx#asahi-dev · 2023-01-25 15:05
axboefor?#asahi-dev · 2023-01-25 15:03
axboejust realized we didn't have crc32c enabled on osx for arm64...#asahi-dev · 2023-01-25 15:03
axboehttps://git.kernel.dk/cgit/fio/commit/?id=1e8ec88fd5f3ab4b7bbd0119708d94fd64a4e7ad#asahi-dev · 2023-01-25 14:53
axboeRy_Darcy_: have your job file too? curious on the m2 max for comparison#asahi-dev · 2023-01-18 15:22
axboesven: nice!!#asahi-dev · 2023-01-04 15:19
axboemarcan: works ;)#asahi-dev · 2022-03-21 17:05
axboesven: thanks for writing an actual commit message :)#asahi-dev · 2022-03-18 22:35
axboelooks at 13 branches to rebase, eek#asahi-dev · 2022-03-18 22:35
axboeI may just rebase your branches on current git + current linux-next for some localized testing#asahi-dev · 2022-03-18 22:34
axboemakes sense#asahi-dev · 2022-03-18 22:31
axboeplan is to fwd when 5.17 is released, to 5.17 + linux-next?#asahi-dev · 2022-03-18 22:31
axboekeep waiting to update, but it's on that base still#asahi-dev · 2022-03-18 22:31
axboemarcan: any reason why you're sticking with the 8+ day old linux-next?#asahi-dev · 2022-03-18 16:04
axboebeen working on non-m1 things on the m1, still works as daily just fine#asahi-dev · 2022-03-18 16:03
axboebtw, feel free to email about issues like that, depending on what I'm working on, I may not drop in here for days. in fact it has been days, just got lucky ;)#asahi-dev · 2022-03-18 16:03
axboeeither that, or make the flush conditional, but seems cleaner to just always init it#asahi-dev · 2022-03-18 16:02
axboetpw_rules: that INIT_DELAYED_WORK() in apple_nvme_probe() should be outside of the if (flush_interval) {} section#asahi-dev · 2022-03-18 16:01
axboetpw_rules: it's using an older one, I did fix the always-init#asahi-dev · 2022-03-18 16:00
axboeit does, iirc I made that explicit#asahi-dev · 2022-03-01 18:18
axboetpw_rules: understood :)#asahi-dev · 2022-03-01 18:17
axboetpw_rules: should just make it use the dynamic handler which exists on macos now#asahi-dev · 2022-03-01 18:16
axboeor you know, do it once at startup#asahi-dev · 2022-03-01 18:16
axboenot sure how often they call GetPageSize, but seems misguided to hardcode it when you can just get the damn page size#asahi-dev · 2022-03-01 18:16
axboefor reference, here's the one I referred to#asahi-dev · 2022-03-01 18:16
axboehttps://chromium-review.googlesource.com/c/chromium/src/+/2361202/#asahi-dev · 2022-03-01 18:15
axboenope#asahi-dev · 2022-03-01 18:15
axboetpw_rules: I might take a stab at it, if someone else can test it. I don't personally care about chrome, but I know others do because of dependencies#asahi-dev · 2022-03-01 18:15
axboealyssa: I preferred to not find out...#asahi-dev · 2022-03-01 18:11
axboethey did provide a reference to the macos != 4k patch, so should be trivial. but not sure I want to dive into compiling chromium locally to test the minor change needed#asahi-dev · 2022-03-01 18:11
axboechromium said they'd "take a patch, but not going to work on it"...#asahi-dev · 2022-02-25 15:13
axboeand figured employer could sponsor another development laptop :)#asahi-dev · 2022-02-25 15:13
axboeI got max envy when I saw those compile numbers