Frühere Suchanfragen

Keine früheren Suchanfragen

Suchoptionen

Nur verfügbar, wenn angemeldet.
mastodontech.de ist einer von vielen unabhängigen Mastodon-Servern, mit dem du dich im Fediverse beteiligen kannst.
Offen für alle (über 16) und bereitgestellt von Markus'Blog

Verwaltet von:

Serverstatistik:

1,4 Tsd.
aktive Profile

mastodontech.de: ÜberStatusProfilverzeichnisDatenschutzerklärung

Mastodon: ÜberApp herunterladenTastenkombinationenQuellcode anzeigenv4.4.1

#else

1 Beitrag1 Beteiligte*r0 Beiträge heute
Fortgeführter Thread
witch_t *navi

one key takeway i had from this soname / abi != package-version / api is that adding fields to structures shouldn’t be a thing to worry about that much, removing them yes, adding, reordering, etc, no

libraries shouldn’t keep their data “obscured” behind a malloc-d pointer just for the sake of binary compatibility. there is other reasons for opaque pointers, but more often than not, a library will be better by letting users choose where to allocate data, be it on the stack or on the heap or statically or so on

and if you worry about people accessing fields they shouldn’t, c23 provides [[deprecated]], meaning you can do:

#if !defined(MYLIB_INTERNAL) && __STDC_VERSION__ >= 202311L
#    define mylib_private [[deprecated("private struct member, do not use.")]]
#else
#    define mylib_private
#endif

and mark struct members as such

distros should have mechanism to easily trigger rebuilds on soname changes, as well as mechanisms to keep the old soname around as long as there’s packages linked to it

the fact that some distros don’t, is not a limitation of the system, but an implementation failure

Antwortete im Thread
Siguza

@ranvel to be more specific, for a given folder PREFIX, this is my ncurses config:

./configure --prefix="$PREFIX" --enable-widec

and this is my nano config:

./configure --prefix="$PREFIX" --disable-nls --enable-utf8 --enable-year2038 NCURSESW_CFLAGS="-I$PREFIX/include" NCURSESW_LIBS="-L$PREFIX/lib -lncursesw"

With ncurses 6.5 and nano 8.5, that leads to a bunch of linker failures for imports with "$NCURSES60" suffix... which comes from Apple's SDK. And if you look at nano src, you can find this:

/* Prefer wide ncurses over normal ncurses over curses. */
#if defined(HAVE_NCURSESW_NCURSES_H)
#include <ncursesw/ncurses.h>
#elif defined(HAVE_NCURSES_H)
#include <ncurses.h>
#else
#include <curses.h>
#endif

So unless you define HAVE_NCURSESW_NCURSES_H, it will include the wrong header. And while HAVE_NCURSES_H shows up in configure, HAVE_NCURSESW_NCURSES_H does not...

#if#include#elif
Antwortete im Thread
Thierry Laurion

@libreleah @mkukri

Both Heads and lbmk permit to apply patches on top of a coreboot fork.

The difference between the two here is that lbmk builds the tree, clean, for each boards, where Heads applies the patches to a fork once, and each board reuses fork build artifacts;, building board specifics in a board specific artifact directory. That permits crossgcc, being the buildstack of each coreboot fork version to be built once, and also repro build issues upstream, economizing both disk space, cpu resource for user and CI.

In Heads goal of building fully functional roms, CI can build and stitch reproducible roms for each commit for end users to download directly from CI, for each commit, and see if a comit broke a built, for each commit. CI cache is reused, so that we don't waste CI resources either.

In the case of t480, the patch was made with lbmk in mind, not coreboot nor Heads, and breaks other thinkpads in coreboot upstream, trying to not only build for t480 but make sure t480 patchset doesn't break other boards. In this case, it breaks all other thinkpads, so prevent Heads from merging the PR. What you propose here is for libreboot and Heads to maintain a patchset not merged upstream; it might suit libreboot mindset, being more bleeding edge, and minifree, selling the t480, but not Heads. Heads tries to stay as close as possible to upstream forks, and pushes upstream projects to merge patches. Its long, not easy, but the right thing to do. The patches stays in a patch dir for everyone to see, per software version. In this case, patches/coreboot-24.12/*

I tried to apply the following patch without success instead of commenting thermal.asl

+diff --git a/src/ec/lenovo/h8/acpi/ec.asl b/src/ec/lenovo/h8/acpi/ec.asl
+index bc54d3b..a0408c8 100644
+--- a/src/ec/lenovo/h8/acpi/ec.asl
++++ b/src/ec/lenovo/h8/acpi/ec.asl
+@@ -331,7 +331,13 @@ Device(EC)
+ #include "sleepbutton.asl"
+ #include "lid.asl"
+ #include "beep.asl"
++
++#ifndef CONFIG_BOARD_LENOVO_T480
+ #include "thermal.asl"
++#else
++//#include "thermal.asl"
++#endif
++
+ #include "systemstatus.asl"
+ #include "thinkpad.asl"
+ }

Other non t480 fail to build, and I have no more time to spend on this. The community is interested, tried to reach libreboot and were seen as spammers.

Please fix your patchset upstream. People saw the t480 being "supported by coreboot" in a talk. People didn't understand it was a WiP patchset under coreboot. And here we are. 24.12 was december 2024 "release", there will be another one in 25.03... I do not have time to maintain patches on top of patches, Leah. My focus is not to be a coreboot distribution. My focus is to deliver reproducible roms to users needing accessible security, and improve that UX. There is no grub/seabios under Heads, my focus is to make upstream do the right thing and participate upstream, and make contributors participate upstream. Here, you stated loud and clear tha libreboot comes first before coreboot, I respect that. But the t480 patchset is the one too from upstream. That upstream patch needs to build, and then will be merged and then you won't have to maintain it either. And others will fix audio issues, nvidia etc. Otherwise its silo work, and i'm not interested in that anymore

---

Yes, there is different coreboot forks specified in a central place: modules/coreboot.

And there, the buildsystem says if it can reuse crossgcc of another fork to fasten builds for each commit. The idea here is that the user building one board, or multiple boards will get the same result, but CI building multiple boards based on the same fork will speed up builds massively.

d16 will move to fam15h fork from other community effort. I mentor now, I don't try to do everything myself. Just as here, trying tto collaborate with you so you fix what was brought up upstream. But up to now, you are upstream for t480.

The goal here was not to compare our buildsystems, simply stating that the patchset upstream will never be merged if it causes regressions building other boards. Libreboot can do what it wants, but needs to respect how coreboot works. Their CI does the same, and make sure that building a commit for a board won't break others. In current case, it breaks others and needs to be updated.

This needs to be fixed upstream at review.coreboot.org/c/coreboot

Julian Andres Klode 🏳️‍🌈

GCC is so annoying, the error attribute can only be set on functions, but I want to set it on namespaces and types too.

Basically I want to define

```
#ifdef APT_COMPILING_APT
#define APT_INTERNAL
#else
#define APT_INTERNAL __attribute__((error("Internal use only")))
#endif
```

Such that external users can't use internal types and functions and namespaces that are in the header, without me having to wrap them in ifdef

AngesagtLive-Feeds
Über