From mboxrd@z Thu Jan 1 00:00:00 1970 Subject: Re: [PATCH v3] rust: add rust-analyzer to the SDK behind a PACKAGECONFIG To: openembedded-core@lists.openembedded.org From: "Dora, Sunil Kumar" X-Originating-Location: Bengaluru, Karnataka, IN (106.51.222.193) X-Originating-Platform: Windows Chrome 157 User-Agent: GROUPS.IO Web Poster MIME-Version: 1.0 Date: Thu, 12 Nov 2026 23:57:34 -0800 References: <20261006111230.1478039-1-benhur.tessele@est.tech> <20261111154442.241009-1-benhur.tessele@est.tech> <743327.1794550799903849632@lists.openembedded.org> In-Reply-To: Message-ID: <727417.1794556654997095405@lists.openembedded.org> Content-Type: multipart/alternative; boundary="NpPMc8ZRhZ9zLe4kskHP" List-Id: X-Webhook-Received: from 45-33-107-173.ip.linodeusercontent.com [45.33.107.173] by lin-tor-korg-lkml-alma10-1 with HTTPS for ; Fri, 13 Nov 2026 07:57:40 -0000 X-Groupsio-URL: https://lists.openembedded.org/g/openembedded-core/message/250153 --NpPMc8ZRhZ9zLe4kskHP Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable On Fri, Nov 13, 2026 at 12:17 PM, Khem Raj wrote: >=20 > binary is started by the system loader, but then picks up a libc.so.6 fro= m > some other glibc tree > (the.../lib/ in the message). That libc wants a private symbol the system > loader doesn=E2=80=99t > provide, so rustc dies before doing anything. >=20 Yes. This looks like the same issue as YOCTO #15579 / #15735. Two patches to drop the wrapper were posted in early 2025 and not merged, the concern being the libncurses case the wrapper was added for in 388e7cac= 9f ( https://git.openembedded.org/openembedded-core/commit/?id=3D388e7cac9f= 90e79ce8c3c1683d8ee0f4df1bc907 ). What has changed since then is glibc 2.44: the pointer guard moved to a GLIBC_PRIVATE symbol exported by ld.so, so any host loader older than 2.44 now fails deterministically on the SDK libc: https://sourceware.org/git/?p=3Dglibc.git;a=3Dcommit;h=3Da5ec880f808ee7268d= 985bed4f961799bdc0a4bf With rust-analyzer in the SDK this will be hit by anyone who has rustup installed, so it may be worth revisiting. How would you like this tracked? Thanks, Sunil Dora >=20 >=20 > --NpPMc8ZRhZ9zLe4kskHP Content-Type: text/html; charset="utf-8" Content-Transfer-Encoding: quoted-printable
On Fri, Nov 13, 2026 at 12:17 PM, Khem Raj wrote:
binary is started by the system loader, but then picks up a libc.so.6 from some other glibc tree
(the .../lib/ in the message). That libc want= s a private symbol the system loader doesn’t
provide, so rustc dies before doing anything.
Yes. This looks like the same issue as YOCTO #15579 / #15735. 
Tw= o patches to drop the wrapper were posted in early 2025 and not merged,&nbs= p;
the concern being the libncurses case the wrapper was added for in = 38= 8e7cac9f.

What has changed since then is glibc 2.44: the poi= nter guard moved to a
GLIBC_PRIVATE symbol exported by ld.so, so any h= ost loader older than
2.44 now fails deterministically on the SDK libc= :

https://sourceware.org/git/?p=3Dglibc.git;a=3Dcommit;h=3Da5ec8= 80f808ee7268d985bed4f961799bdc0a4bf

With rust-analyzer in the SD= K this will be hit by anyone who has rustup
installed, so it may be wo= rth revisiting. How would you like this tracked?

Thanks,
Su= nil Dora
 
--NpPMc8ZRhZ9zLe4kskHP--