Vitali Lovich

~~Tracking it down further, it has to be a write of 4096 to position 0 followed by an equivalent read. If I write 8192 for example, this issue doesn't repro....

Here's some test cases of initial write followed by a read. No pattern is immediately discernible for me. The last two test cases in particular seem fairly weird... |write pos|write...

https://gist.github.com/vlovich/9ab1bbb9b7c6081365c49b2ce35bf713 works. Not sure if that maps 1:1 with what's happening in glommio, but I have to imagine it's pretty close. @glommer does the C code look correct with respect...

I printed all the buffers used by `sqe.prep_read` and the pointer is always 4096 aligned.

Instrumented the kernel to get more insight. The line that's returning an error is this: https://elixir.bootlin.com/linux/v6.5.5/source/block/blk-core.c#L732 ~~I did disable the read/write queues as per https://blog.cloudflare.com/speeding-up-linux-disk-encryption/ / https://wiki.archlinux.org/title/Dm-crypt/Specialties#Disable_workqueue_for_increased_solid_state_drive_(SSD)_performance which I'm assuming...

Nope, Rc isn't aliased to anything. I already have covariance specified on another field so the macro fails with `TODO: Nice error, covariance specified twice` if I also add either...

Is this related at all to #3223 ?

I *think* we need something like this. Basically the problem we have is that we apply the template in a loop for some new variables as they arrive. However, it's...

Not to be "that guy" but it works for me even without `SWCRC=true`. ``` $ cat .swcrc { "env": { "debug": true } } ``` This causes a whole bunch...

Oh interesting. But it doesn't work if I'm using `@swc-node/register/esm` as the `-r` flag. So I wonder if you're in a similar code path.