[ changing the subject for this old mail from last month, I've
only just had time to retry the rustc build ]
On Tue, Dec 25, 2018 at 09:07:57PM +0200, Thanos Baloukas via blfs-dev wrote:
> On 25/12/18 7:55 π.μ., Ken Moffat via blfs-dev wrote:
> > So, for my attempts to keep old systems viable, things are looking
> > better, at least until firefox-65 appears (that will need a newer
> > version of rustc, but for me 1.31.1 fails to build from source).
> >
> Well, rust really hates you! ):
>
[ trimming a lot of this ]
>
> Note that I've never installed libssh2. Rust downloads it, so for me
> it should not be required. 1.31.0 had downloaded libssh2-sys v0.2.11
> that 1.31.1 used as well, and on my previous system rustc-1.27.0 had
> downloaded v0.2.6
I think I replied to that part last month - the download is a cargo
file, I suspect it links to system libssh2.
>
> On my llvm builds I add the -DLLVM_ENABLE_RTTI=ON option because mesa
> with meson needs it IIRC.
Keeping that comment as a reminder for whoever looks at meson.
>
> I don't use python3. I do
>
> ./x.py build
> DESTDIR=${fakeroot} ./x.py install
>
> x.py has #!/usr/bin/env python
As I think I already said, I had noticed that. It worked in 1.29.2,
but it does seem to be the reason why my build of 1.31.1 failed,
which I'll now show for the archives and for gurgle:
---
Building stage1 std artifacts (x86_64-unknown-linux-gnu ->
x86_64-unknown-linux-gnu)
error: failed to run `rustc` to learn about target-specific information
Caused by:
process didn't exit successfully:
`/scratch/ken/rustc-1.31.1-src/build/bootstrap/debug/rustc - --crate-name ___
--print=file-names -C link-args=-lffi --target x86_64-unknown-linux-gnu
--crate-type bin --crate-type rlib --crate-type dylib --crate-type cdylib
--crate-type staticlib --crate-type proc-macro` (signal: 11, SIGSEGV: invalid
memory reference)
command did not execute successfully:
"/scratch/ken/rustc-1.31.1-src/build/x86_64-unknown-linux-gnu/stage0/bin/cargo"
"build" "--target" "x86_64-unknown-linux-gnu" "-j" "4" "--release" "--features"
"panic-unwind jemalloc backtrace" "--manifest-path"
"/scratch/ken/rustc-1.31.1-src/src/libstd/Cargo.toml" "--message-format" "json"
expected success, got: exit code: 101
thread 'main' panicked at 'cargo must succeed', bootstrap/compile.rs:1115:9
note: Run with `RUST_BACKTRACE=1` for a backtrace.
failed to run: /scratch/ken/rustc-1.31.1-src/build/bootstrap/debug/bootstrap
build
Build completed unsuccessfully in 0:00:02
---
Using python2, I successfully completed the build a little while
ago, but failed to notice my own config.toml contained a variable
which I had not set for the manual build (the docdir), causing the
DESTDIR install to fail.
Doh! Didn't actually send this. Manual build, DESTDIR, tests now
completed. On ryzen, there are now TEN reports of invalid opcodes,
and 14 test failures. On the bright side, the totals for run,
passed, failed, ignored are now only 3 adrift - maybe the rustaceans
and pythonistas will eventually manage to get their maths working
;-)
In 1.31.1, the three thumb test failures (because we only build x86)
now appear twice (maybe they already did in 1.29), and as well as the
seven failures for gdb-8.1 there is one new failure:
[run-make] run-make-fulldeps/sysroot-crates-are-unstable
which only reports:
No such file or directory: 'lib/rustlib/x86_64-unknown-linux-gnu/lib'
I'll need to remember to change the book to use python2 for the
build, tests, and DESTDIR installs, but not touching that until
after I've stretched my luck and actually done a scripted install
with tests.
Thanks again for your help.
ĸen
--
Take three of these a minute for four minutes. Don't take with
alcohol or you'll grow an extra head.
-- The Doctor
--
http://lists.linuxfromscratch.org/listinfo/blfs-dev
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page
