| 13 Jan 2022 |
 Zhaofeng Li | So the http2 codepath does use the correct parsed representation for the content length, and only http3 does this crap. | 20:03:36 |
 woobilicious | I wonder why Find() hasn't been implemented to always be case insensitive...seems like an awful waste of mental effort to remember to apply aIgnoreCase every time you do a search. | 22:07:23 |
| woobilicious | Oh right, mFlatHttpRequestHeaders is a string, it's not a header store class. | 22:14:37 |
| woobilicious | I have a kernel thread that's acting weird, how do I inspect the stack? | 22:49:57 |
| Zhaofeng Li | In reply to @woobilicious:matrix.org
I have a kernel thread that's acting weird, how do I inspect the stack? thread apply all backtrace. Define an alias in your .gdbinit
define btall
thread apply all backtrace
end
| 23:09:20 |
| Zhaofeng Li | In reply to @woobilicious:matrix.org
I wonder why Find() hasn't been implemented to always be case insensitive...seems like an awful waste of mental effort to remember to apply aIgnoreCase every time you do a search. Why should it? Uppercase isn't the same thing as lowercase. In this specific case it is, but case sensitivity is the least of their worries here | 23:10:25 |
| woobilicious | In reply to @zhaofeng:zhaofeng.li
Why should it? Uppercase isn't the same thing as lowercase. In this specific case it is, but case sensitivity is the least of their worries here As I said earlier, I thought that was a specific object for handling headers, not just a generic string search. In other languages I've used libraries that implement headers as a hashmap. | 23:22:57 |
| 15 Jan 2022 |
 Mic92 (Old) | In reply to @woobilicious:matrix.org
I have a kernel thread that's acting weird, how do I inspect the stack? If you know the pid of your kernel thread: sudo cat /proc/162/stack
[<0>] taskq_thread+0x3b7/0x480 [spl]
[<0>] kthread+0x124/0x150
[<0>] ret_from_fork+0x1f/0x30 | 06:49:06 |
| woobilicious | Ahh thanks Mic92 ! | 07:31:49 |
| 16 Jan 2022 |
|  Andreas Schrägle joined the room. | 02:08:01 |
| 18 Jan 2022 |
 Qubasa | I just found out in the Linux kernel that for any source file kernel/foo.c you can run "make kernel/foo.i" to get the preprocessed source | 10:37:04 |
 Finn Behrens (he/him) | That is a nice feature | 10:40:29 |
| Finn Behrens (he/him) | May I cite you in the rust-for-Linux Zulip? | 10:40:59 |
| Qubasa | Sure ^^ | 11:27:18 |
| Finn Behrens (he/him) | Done, let's hope they agree to create the same feature for rust files :) | 11:30:34 |
| Mic92 (Old) | I thought this works for make in general? | 19:47:28 |
| Zhaofeng Li | They would have to define rules to run the preprocessor (or rustc expansion) separately | 19:59:50 |
| Finn Behrens (he/him) | Rust-for-Linux wants it: https://github.com/Rust-for-Linux/linux/issues/633 | 20:54:38 |
| 20 Jan 2022 |
| Mic92 (Old) | Yesterday I learned about strace -yy: Gives you more context for file descriptors | 11:48:21 |
| 21 Jan 2022 |
| Qubasa | A benchmarking tool that can measure cycle execution time of single instructions: https://github.com/andreas-abel/nanoBench#cycle-by-cycle-measurements | 07:26:30 |
|  Emantor joined the room. | 13:04:05 |
| 4 Feb 2022 |
| Mic92 (Old) | autopatchelfhook should be now a lot faster: https://github.com/NixOS/nixpkgs/pull/149731/files | 13:13:10 |
| Mic92 (Old) | In reply to @qubasa:gchq.icu
A benchmarking tool that can measure cycle execution time of single instructions: https://github.com/andreas-abel/nanoBench#cycle-by-cycle-measurements When would you use that? | 13:14:21 |
 K900 ⚡️ | In reply to @mic92:nixos.dev
autopatchelfhook should be now a lot faster: https://github.com/NixOS/nixpkgs/pull/149731/files Had this idea for a while now, why not expose auto-patchelf as a separate script too? | 13:22:10 |
| K900 ⚡️ | Could be nice for running external binaries | 13:22:21 |
| Mic92 (Old) | @k900 I guess so far no one wanted to make this a stable api... | 13:23:07 |
| Mic92 (Old) | How would you use it on external binaries? i.e. where do you get your libraries from? | 13:23:27 |
 andi- | CLI arguments? IMO the script should not use random environment variables but be called explictly with the arguments defined. | 13:28:12 |
| andi- | and then "auto" part that is in Nixpkgs could probably do that ENV -> CLI arguments part for us | 13:28:43 |
| K900 ⚡️ | Yeah, I'm assuming something like auto-patchelf ./some-bin nixpkgs#foo nixpkgs#bar or something like that | 13:37:28 |
