Re: [OE-core] [RFC 1/5] valgrind: make ptest depend on all components

Re: [OE-core] [RFC 1/5] valgrind: make ptest depend on all components

[Randy MacLeod]

[-- Attachment #1: Type: text/plain, Size: 3367 bytes --]

On 2023-11-08 8:25 p.m., Randy MacLeod via lists.openembedded.org wrote:
> From: Randy MacLeod<Randy.MacLeod@windriver.com>
>
> When the helper scripts were split into separate packages, they
> weren't added as ptest dependencies. Fix that.

Well, this is embarrassing...

I'm no longer convinced that this is required.

I think I saw the ../cg_diff errors below:

PASS: memcheck/tests/vcpu_fbench
PASS: memcheck/tests/wcs
PASS: memcheck/tests/wcsncpy
PASS: memcheck/tests/wmemcmp
FAIL: memcheck/tests/wrap1
FAIL: memcheck/tests/wrap2
FAIL: memcheck/tests/wrap3
FAIL: memcheck/tests/wrap4
FAIL: memcheck/tests/wrap5
FAIL: memcheck/tests/wrap6
FAIL: memcheck/tests/wrap7
FAIL: memcheck/tests/wrap8
PASS: memcheck/tests/wrapmalloc
PASS: memcheck/tests/wrapmallocstatic
PASS: memcheck/tests/writev1
PASS: memcheck/tests/xml1
python3: can't open file 
'/usr/lib/valgrind/ptest/cachegrind/tests/../cg_diff': [Errno 2] No such 
file or directory
post check failed: python3 ../cg_diff --mod-funcname="s/main/MAIN/" 
ann1.cgout ann1b.cgout > ann-diff1.cgout && python3 ../cg_annotate 
--mod-filename="s/a.c/A.c/" --mod-funcname s/MAIN/Main/
PASS: cachegrind/tests/ann-diff1
python3: can't open file 
'/usr/lib/valgrind/ptest/cachegrind/tests/../cg_diff': [Errno 2] No such 
file or directory
post check failed: python3 ../cg_diff --mod-filename="s/.*aux\//aux\//i" 
--mod-funcname="s/(f[a-z]*)[0-9]/\1N/g" ann-diff2a.cgout 
ann-diff2b.cgout > ann-diff2c.cgout && python3 ../cg_annotat
PASS: cachegrind/tests/ann-diff2

and made an assumption that may not be valid. I think this problem is 
still present
even after this change.

As I said previously, I may not have time to get back to this until next 
week so
if anyone picks up this work, they might want to confirm that this is 
really required.

I should have just waited until I had time to complete the update before 
spamming the list.

../Randy

../Randy


>
> Signed-off-by: Randy MacLeod<Randy.MacLeod@windriver.com>
> ---
>   meta/recipes-devtools/valgrind/valgrind_3.21.0.bb | 4 +++-
>   1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/meta/recipes-devtools/valgrind/valgrind_3.21.0.bb b/meta/recipes-devtools/valgrind/valgrind_3.21.0.bb
> index 68ab319a1b..a2feda6aae 100644
> --- a/meta/recipes-devtools/valgrind/valgrind_3.21.0.bb
> +++ b/meta/recipes-devtools/valgrind/valgrind_3.21.0.bb
> @@ -139,7 +139,9 @@ RDEPENDS:${PN}-ptest += " bash coreutils curl file \
>      perl-module-overloading perl-module-cwd perl-module-ipc-open3 \
>      perl-module-carp perl-module-symbol \
>      procps sed ${PN}-dbg ${PN}-src ${TCLIBC}-src gcc-runtime-dbg \
> -   util-linux-taskset"
> +   util-linux-taskset \
> +   ${PN}-cachegrind ${PN}-massif ${PN}-callgrind \
> +"
>   RDEPENDS:${PN}-ptest:append:libc-glibc = " glibc-utils glibc-gconv-utf-32"
>   
>   # One of the tests contains a bogus interpreter path on purpose.
>
> -=-=-=-=-=-=-=-=-=-=-=-
> Links: You receive all messages sent to this group.
> View/Reply Online (#190367):https://lists.openembedded.org/g/openembedded-core/message/190367
> Mute This Topic:https://lists.openembedded.org/mt/102477896/3616765
> Group Owner:openembedded-core+owner@lists.openembedded.org
> Unsubscribe:https://lists.openembedded.org/g/openembedded-core/unsub  [randy.macleod@windriver.com]
> -=-=-=-=-=-=-=-=-=-=-=-
>

-- 
# Randy MacLeod
# Wind River Linux

[-- Attachment #2: Type: text/html, Size: 5614 bytes --]

Re: [OE-core] [RFC 1/5] valgrind: make ptest depend on all components

[Alexander Kanavin]

On Thu, 9 Nov 2023 at 03:00, Randy MacLeod <randy.macleod@windriver.com> wrote:
> As I said previously, I may not have time to get back to this until next week so
> if anyone picks up this work, they might want to confirm that this is really required.
>
> I should have just waited until I had time to complete the update before spamming the list.

Thanks Randy, I hope you find time to complete this, as I'm planning
to work on other updates. We're falling behind all over the place, as
seen in the latest AUH report. Rust would be nice to do too.

Alex

Re: [OE-core] [RFC 1/5] valgrind: make ptest depend on all components

[Randy MacLeod]

[-- Attachment #1: Type: text/plain, Size: 1078 bytes --]

On 2023-11-15 9:48 a.m., Alexander Kanavin wrote:
> On Thu, 9 Nov 2023 at 03:00, Randy MacLeod<randy.macleod@windriver.com>  wrote:
>> As I said previously, I may not have time to get back to this until next week so
>> if anyone picks up this work, they might want to confirm that this is really required.
>>
>> I should have just waited until I had time to complete the update before spamming the list.
> Thanks Randy, I hope you find time to complete this, as I'm planning
> to work on other updates.
Yes, I will get to it but I have some things happening outside of work
that will make this take longer than usual. I did make a bit more progress
on fixing the ptest regressions.
> We're falling behind all over the place, as
> seen in the latest AUH report. Rust would be nice to do too.

Indeed.

Yash and Sundeep have 1.72/1.73 updated and are working on

getting the test suite to work but struggling with that.
I asked him to track what's happening in:
https://bugzilla.yoctoproject.org/show_bug.cgi?id=15275

../Randy


>
> Alex


-- 
# Randy MacLeod
# Wind River Linux

[-- Attachment #2: Type: text/html, Size: 2306 bytes --]

Re: [OE-core] [RFC 1/5] valgrind: make ptest depend on all components

[Alex Kiernan]

On Wed, Nov 15, 2023 at 4:30 PM Randy MacLeod via
lists.openembedded.org
<randy.macleod=windriver.com@lists.openembedded.org> wrote:
>
> On 2023-11-15 9:48 a.m., Alexander Kanavin wrote:
>
...
> Rust would be nice to do too.
>
> Indeed.
>
> Yash and Sundeep have 1.72/1.73 updated and are working on
>
> getting the test suite to work but struggling with that.
> I asked him to track what's happening in:
>    https://bugzilla.yoctoproject.org/show_bug.cgi?id=15275
>

I did start looking at this... mostly I've failed to find enough time
to even get to the point where I've got a useful environment for
running the tests outside of bitbake (and hence avoid a complete rust
build cycle).

Rust 1.74.0 landed yesterday, so I'm going to have another look... if
it's useful to anyone my (very much WIP) tree is here:

https://github.com/akiernan/poky


--
Alex Kiernan

Re: [OE-core] [RFC 1/5] valgrind: make ptest depend on all components

[Alex Kiernan]

On Fri, Nov 17, 2023 at 12:04 PM Alex Kiernan via
lists.openembedded.org <alex.kiernan=gmail.com@lists.openembedded.org>
wrote:
>
> On Wed, Nov 15, 2023 at 4:30 PM Randy MacLeod via
> lists.openembedded.org
> <randy.macleod=windriver.com@lists.openembedded.org> wrote:
> >
> > On 2023-11-15 9:48 a.m., Alexander Kanavin wrote:
> >
> ...
> > Rust would be nice to do too.
> >
> > Indeed.
> >
> > Yash and Sundeep have 1.72/1.73 updated and are working on
> >
> > getting the test suite to work but struggling with that.
> > I asked him to track what's happening in:
> >    https://bugzilla.yoctoproject.org/show_bug.cgi?id=15275
> >
>
> I did start looking at this... mostly I've failed to find enough time
> to even get to the point where I've got a useful environment for
> running the tests outside of bitbake (and hence avoid a complete rust
> build cycle).
>
> Rust 1.74.0 landed yesterday, so I'm going to have another look... if
> it's useful to anyone my (very much WIP) tree is here:
>
> https://github.com/akiernan/poky
>

I've finally got this running in a way I can usefully inspect and try
things and it's beginning to feel like we've headed into something
which simply isn't supported.

The failure is this:

  thread 'main' panicked at src/tools/compiletest/src/common.rs:519:30:
  no entry found for key

Which is empty because the mapping list is collected by running rustc
with -Z unstable-options --print=all-target-specs-json, but that fails
because we're not running a nightly compiler:

  error: the option `Z` is only accepted on the nightly compiler
  help: consider switching to a nightly toolchain: `rustup default nightly`
  note: selecting a toolchain with `+toolchain` arguments require a
rustup proxy; see
<https://rust-lang.github.io/rustup/concepts/index.html>
  note: for more information about Rust's stability policy, see
<https://doc.rust-lang.org/book/appendix-07-nightly-rust.html#unstable-features>

Basically feels like running the test suite is only supported on the
nightlies...

This is the full piece of useful logs for reference:

Testing stage1 compiletest suite=run-coverage mode=run-coverage
(aarch64-unknown-linux-gnu -> aarch64-poky-linux-gnu)
thread 'main' panicked at src/tools/compiletest/src/common.rs:519:30:
no entry found for key
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
Note that not all variants of mir-opt tests are going to be blessed,
as no mapping between
a 32bit and a 64bit target was found for aarch64-poky-linux-gnu.
You can add that mapping by changing MIR_OPT_BLESS_TARGET_MAPPING in
src/bootstrap/test.rs
Testing stage1 compiletest suite=mir-opt mode=mir-opt
(aarch64-unknown-linux-gnu -> aarch64-poky-linux-gnu)
thread 'main' panicked at src/tools/compiletest/src/common.rs:519:30:
no entry found for key
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
error: the option `Z` is only accepted on the nightly compiler

help: consider switching to a nightly toolchain: `rustup default nightly`

note: selecting a toolchain with `+toolchain` arguments require a
rustup proxy; see
<https://rust-lang.github.io/rustup/concepts/index.html>

note: for more information about Rust's stability policy, see
<https://doc.rust-lang.org/book/appendix-07-nightly-rust.html#unstable-features>

error: 1 nightly option were parsed

thread 'main' panicked at synthetic_targets.rs:66:9:
failed to gather the target spec for aarch64-poky-linux-gnu
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

-- 
Alex Kiernan

Re: [OE-core] [RFC 1/5] valgrind: make ptest depend on all components

[Yash Shinde]

On 20-11-2023 21:57, Alex Kiernan wrote:
> CAUTION: This email comes from a non Wind River email account!
> Do not click links or open attachments unless you recognize the sender and know the content is safe.
>
> I've finally got this running in a way I can usefully inspect and try
> things and it's beginning to feel like we've headed into something
> which simply isn't supported.
>
> The failure is this:
>
>    thread 'main' panicked at src/tools/compiletest/src/common.rs:519:30:
>    no entry found for key
>
> Which is empty because the mapping list is collected by running rustc
> with -Z unstable-options --print=all-target-specs-json, but that fails
> because we're not running a nightly compiler:
>
>    error: the option `Z` is only accepted on the nightly compiler
>    help: consider switching to a nightly toolchain: `rustup default nightly`
>    note: selecting a toolchain with `+toolchain` arguments require a
> rustup proxy; see
> <https://rust-lang.github.io/rustup/concepts/index.html>
>    note: for more information about Rust's stability policy, see
> <https://doc.rust-lang.org/book/appendix-07-nightly-rust.html#unstable-features>
>
> Basically feels like running the test suite is only supported on the
> nightlies...
I saw the same error earlier.

Investigating further, we understood that the issue is not related to 
running on nightly channel. The '-Z unstable options' were also present 
in earlier rust versions and removing of the newly added -Z options 
after 1.70 also didn't solve the problem.

On doing further analysis, we found that the custom target config/specs 
were not fetched during the rust build.

The log for reference is as follows:

Building stage1 library artifacts (x86_64-unknown-linux-gnu -> 
x86_64-poky-linux-gnu)
error: failed to run `rustc` to learn about target-specific information

Caused by:
   process didn't exit successfully: ` 
build-st/tmp/work/core2-64-poky-linux/rust/1.72.0/rustc-1.72.0-src/build/bootstrap/debug/rustc 
- --crate-name ___ --print=file-names -Csymbol-mangling-version=legacy 
-Zunstable-options -Zunstable-options '--check-cfg=values(bootstrap)' 
'--check-cfg=values(stdarch_intel_sde)' 
'--check-cfg=values(no_fp_fmt_parse)' 
'--check-cfg=values(no_global_oom_handling)' '--check-cfg=values(no_rc)' 
'--check-cfg=values(no_sync)' '--check-cfg=values(freebsd12)' 
'--check-cfg=values(freebsd13)' 
'--check-cfg=values(backtrace_in_libstd)' 
'--check-cfg=values(target_env,"libnx")' 
'--check-cfg=values(target_arch,"asmjs","spirv","nvptx","xtensa")' 
-Zmacro-backtrace -Clink-args=-Wl,-z,origin 
'-Clink-args=-Wl,-rpath,$ORIGIN/../lib' -Csplit-debuginfo=off 
-Cprefer-dynamic -Zinline-mir -Cembed-bitcode=yes 
'-Zcrate-attr=doc(html_root_url="https://doc.rust-lang.org/1.72.0/")' 
--target x86_64-poky-linux-gnu --crate-type bin --crate-type rlib 
--crate-type dylib --crate-type cdylib --crate-type staticlib 
--crate-type proc-macro --print=sysroot --print=split-debuginfo 
--print=crate-name --print=cfg` (exit status: 1)
   --- stderr
   error: Error loading target specification: Could not find 
specification for target "x86_64-poky-linux-gnu". Run `rustc --print 
target-list` for a list of built-in targets

Build completed unsuccessfully in 0:04:38

Regards,
Yash.

> This is the full piece of useful logs for reference:
>
> Testing stage1 compiletest suite=run-coverage mode=run-coverage
> (aarch64-unknown-linux-gnu -> aarch64-poky-linux-gnu)
> thread 'main' panicked at src/tools/compiletest/src/common.rs:519:30:
> no entry found for key
> note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
> Note that not all variants of mir-opt tests are going to be blessed,
> as no mapping between
> a 32bit and a 64bit target was found for aarch64-poky-linux-gnu.
> You can add that mapping by changing MIR_OPT_BLESS_TARGET_MAPPING in
> src/bootstrap/test.rs
> Testing stage1 compiletest suite=mir-opt mode=mir-opt
> (aarch64-unknown-linux-gnu -> aarch64-poky-linux-gnu)
> thread 'main' panicked at src/tools/compiletest/src/common.rs:519:30:
> no entry found for key
> note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
> error: the option `Z` is only accepted on the nightly compiler
>
> help: consider switching to a nightly toolchain: `rustup default nightly`
>
> note: selecting a toolchain with `+toolchain` arguments require a
> rustup proxy; see
> <https://rust-lang.github.io/rustup/concepts/index.html>
>
> note: for more information about Rust's stability policy, see
> <https://doc.rust-lang.org/book/appendix-07-nightly-rust.html#unstable-features>
>
> error: 1 nightly option were parsed
>
> thread 'main' panicked at synthetic_targets.rs:66:9:
> failed to gather the target spec for aarch64-poky-linux-gnu
> note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
>
> --
> Alex Kiernan

Re: [OE-core] [RFC 1/5] valgrind: make ptest depend on all components

[Alex Kiernan]

On Tue, Nov 21, 2023 at 11:30 AM Yash Shinde <Yash.Shinde@windriver.com> wrote:
>
>
> On 20-11-2023 21:57, Alex Kiernan wrote:
> > CAUTION: This email comes from a non Wind River email account!
> > Do not click links or open attachments unless you recognize the sender and know the content is safe.
> >
> > I've finally got this running in a way I can usefully inspect and try
> > things and it's beginning to feel like we've headed into something
> > which simply isn't supported.
> >
> > The failure is this:
> >
> >    thread 'main' panicked at src/tools/compiletest/src/common.rs:519:30:
> >    no entry found for key
> >
> > Which is empty because the mapping list is collected by running rustc
> > with -Z unstable-options --print=all-target-specs-json, but that fails
> > because we're not running a nightly compiler:
> >
> >    error: the option `Z` is only accepted on the nightly compiler
> >    help: consider switching to a nightly toolchain: `rustup default nightly`
> >    note: selecting a toolchain with `+toolchain` arguments require a
> > rustup proxy; see
> > <https://rust-lang.github.io/rustup/concepts/index.html>
> >    note: for more information about Rust's stability policy, see
> > <https://doc.rust-lang.org/book/appendix-07-nightly-rust.html#unstable-features>
> >
> > Basically feels like running the test suite is only supported on the
> > nightlies...
> I saw the same error earlier.
>
> Investigating further, we understood that the issue is not related to
> running on nightly channel. The '-Z unstable options' were also present
> in earlier rust versions and removing of the newly added -Z options
> after 1.70 also didn't solve the problem.
>

There's a fair number of changes in the test framework around the code
that calls out to rustc with new paths that use
--print=all-target-specs-json unconditionally. Assuming I've read it
right, -Zunstable-options is only evaluated at the point where you
actually try and use such an option which is covered by
unstable-options, so just passing it in doesn't get you a failure.

> On doing further analysis, we found that the custom target config/specs
> were not fetched during the rust build.
>
> The log for reference is as follows:
>
> Building stage1 library artifacts (x86_64-unknown-linux-gnu ->
> x86_64-poky-linux-gnu)
> error: failed to run `rustc` to learn about target-specific information
>
> Caused by:
>    process didn't exit successfully: `
> build-st/tmp/work/core2-64-poky-linux/rust/1.72.0/rustc-1.72.0-src/build/bootstrap/debug/rustc
> - --crate-name ___ --print=file-names -Csymbol-mangling-version=legacy
> -Zunstable-options -Zunstable-options '--check-cfg=values(bootstrap)'
> '--check-cfg=values(stdarch_intel_sde)'
> '--check-cfg=values(no_fp_fmt_parse)'
> '--check-cfg=values(no_global_oom_handling)' '--check-cfg=values(no_rc)'
> '--check-cfg=values(no_sync)' '--check-cfg=values(freebsd12)'
> '--check-cfg=values(freebsd13)'
> '--check-cfg=values(backtrace_in_libstd)'
> '--check-cfg=values(target_env,"libnx")'
> '--check-cfg=values(target_arch,"asmjs","spirv","nvptx","xtensa")'
> -Zmacro-backtrace -Clink-args=-Wl,-z,origin
> '-Clink-args=-Wl,-rpath,$ORIGIN/../lib' -Csplit-debuginfo=off
> -Cprefer-dynamic -Zinline-mir -Cembed-bitcode=yes
> '-Zcrate-attr=doc(html_root_url="https://doc.rust-lang.org/1.72.0/")'
> --target x86_64-poky-linux-gnu --crate-type bin --crate-type rlib
> --crate-type dylib --crate-type cdylib --crate-type staticlib
> --crate-type proc-macro --print=sysroot --print=split-debuginfo
> --print=crate-name --print=cfg` (exit status: 1)
>    --- stderr
>    error: Error loading target specification: Could not find
> specification for target "x86_64-poky-linux-gnu". Run `rustc --print
> target-list` for a list of built-in targets
>

Out of interest, did you try 1.71.0 to see if that failed? Based on
reading the commit log for src/tools/compiletest/src/common.rs, I
suspect it will fail.

-- 
Alex Kiernan

Re: [OE-core] [RFC 1/5] valgrind: make ptest depend on all components

[Yash Shinde]

On 21-11-2023 21:27, Alex Kiernan wrote:
> CAUTION: This email comes from a non Wind River email account!
> Do not click links or open attachments unless you recognize the sender and know the content is safe.
>
> On Tue, Nov 21, 2023 at 11:30 AM Yash Shinde <Yash.Shinde@windriver.com> wrote:
>>
>> On 20-11-2023 21:57, Alex Kiernan wrote:
>>> CAUTION: This email comes from a non Wind River email account!
>>> Do not click links or open attachments unless you recognize the sender and know the content is safe.
>>>
>>> I've finally got this running in a way I can usefully inspect and try
>>> things and it's beginning to feel like we've headed into something
>>> which simply isn't supported.
>>>
>>> The failure is this:
>>>
>>>     thread 'main' panicked at src/tools/compiletest/src/common.rs:519:30:
>>>     no entry found for key
>>>
>>> Which is empty because the mapping list is collected by running rustc
>>> with -Z unstable-options --print=all-target-specs-json, but that fails
>>> because we're not running a nightly compiler:
>>>
>>>     error: the option `Z` is only accepted on the nightly compiler
>>>     help: consider switching to a nightly toolchain: `rustup default nightly`
>>>     note: selecting a toolchain with `+toolchain` arguments require a
>>> rustup proxy; see
>>> <https://rust-lang.github.io/rustup/concepts/index.html>
>>>     note: for more information about Rust's stability policy, see
>>> <https://doc.rust-lang.org/book/appendix-07-nightly-rust.html#unstable-features>
>>>
>>> Basically feels like running the test suite is only supported on the
>>> nightlies...
>> I saw the same error earlier.
>>
>> Investigating further, we understood that the issue is not related to
>> running on nightly channel. The '-Z unstable options' were also present
>> in earlier rust versions and removing of the newly added -Z options
>> after 1.70 also didn't solve the problem.
>>
> There's a fair number of changes in the test framework around the code
> that calls out to rustc with new paths that use
> --print=all-target-specs-json unconditionally. Assuming I've read it
> right, -Zunstable-options is only evaluated at the point where you
> actually try and use such an option which is covered by
> unstable-options, so just passing it in doesn't get you a failure.
>
>> On doing further analysis, we found that the custom target config/specs
>> were not fetched during the rust build.
>>
>> The log for reference is as follows:
>>
>> Building stage1 library artifacts (x86_64-unknown-linux-gnu ->
>> x86_64-poky-linux-gnu)
>> error: failed to run `rustc` to learn about target-specific information
>>
>> Caused by:
>>     process didn't exit successfully: `
>> build-st/tmp/work/core2-64-poky-linux/rust/1.72.0/rustc-1.72.0-src/build/bootstrap/debug/rustc
>> - --crate-name ___ --print=file-names -Csymbol-mangling-version=legacy
>> -Zunstable-options -Zunstable-options '--check-cfg=values(bootstrap)'
>> '--check-cfg=values(stdarch_intel_sde)'
>> '--check-cfg=values(no_fp_fmt_parse)'
>> '--check-cfg=values(no_global_oom_handling)' '--check-cfg=values(no_rc)'
>> '--check-cfg=values(no_sync)' '--check-cfg=values(freebsd12)'
>> '--check-cfg=values(freebsd13)'
>> '--check-cfg=values(backtrace_in_libstd)'
>> '--check-cfg=values(target_env,"libnx")'
>> '--check-cfg=values(target_arch,"asmjs","spirv","nvptx","xtensa")'
>> -Zmacro-backtrace -Clink-args=-Wl,-z,origin
>> '-Clink-args=-Wl,-rpath,$ORIGIN/../lib' -Csplit-debuginfo=off
>> -Cprefer-dynamic -Zinline-mir -Cembed-bitcode=yes
>> '-Zcrate-attr=doc(html_root_url="https://doc.rust-lang.org/1.72.0/")'
>> --target x86_64-poky-linux-gnu --crate-type bin --crate-type rlib
>> --crate-type dylib --crate-type cdylib --crate-type staticlib
>> --crate-type proc-macro --print=sysroot --print=split-debuginfo
>> --print=crate-name --print=cfg` (exit status: 1)
>>     --- stderr
>>     error: Error loading target specification: Could not find
>> specification for target "x86_64-poky-linux-gnu". Run `rustc --print
>> target-list` for a list of built-in targets
>>
> Out of interest, did you try 1.71.0 to see if that failed? Based on
> reading the commit log for src/tools/compiletest/src/common.rs, I
> suspect it will fail.
>
Yes, that fails in 1.71.0 also. The 1.71 build progresses to check some 
test suites before failing as follows-

.........Test cases running from different suites like codegen, ui, 
mir-opt, assembly, incremental....

test [incremental] 
tests/incremental/issue-79661-missing-def-path-hash.rs ... ok
test [incremental] tests/incremental/remapped_paths_cc/main.rs ... ok
test [incremental] tests/incremental/lto.rs ... ok
test result: ok. 161 passed; 0 failed; 2 ignored; 0 measured; 0 filtered 
out; finished in 12.51s
finished in 13.102 seconds
Check compiletest suite=debuginfo mode=debuginfo 
(x86_64-unknown-linux-gnu -> x86_64-poky-linux-gnu)
WARNING: debuginfo tests are not available when testing with remote
running 0 tests

test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 0 filtered 
out; finished in 18.92µs

         finished in 0.147 seconds

Building stage0 library artifacts (x86_64-poky-linux-gnu)
error: process didn't exit successfully: 
`build-st/tmp/work/core2-64-poky-linux/rust/1.71.0/rustc-1.71.0-src/build/bootstrap/debug/rustc 
-vV` (exit status: 101)
--- stderr
thread 'main' panicked at '
Failed to run:
LD_LIBRARY_PATH="/build-st/tmp/work/core2-64-poky-linux/rust/1.71.0/rustc-1.71.0-src/build/x86_64-poky-linux-gnu/stage0-sysroot/lib" 
"build-st/tmp/work/core2-64-poky-linux/rust/1.71.0/rustc-1.71.0-src/build/x86_64-poky-linux-gnu/stage0-sysroot/bin/rustc" 
"-vV" "-Wrust_2018_idioms" "-Wunused_lifetimes" 
"-Wsemicolon_in_expressions_from_macros" "-Dwarnings" 
"-Clinker=build-st/tmp/work/core2-64-poky-linux/rust/1.71.0/wrapper/target-rust-ccld" 
"--cfg=bootstrap" "-Zunstable-options" "--check-cfg=values(bootstrap)" 
"-Z" "force-unstable-if-unmarked"
-------------: Os { code: 2, kind: NotFound, message: "No such file or 
directory" }', bin/rustc.rs:230:37
note: run with `RUST_BACKTRACE=1` environment variable to display a 
backtrace

Build completed unsuccessfully in 0:06:09

Regards,
Yash
> --
> Alex Kiernan

Re: [OE-core] [RFC 1/5] valgrind: make ptest depend on all components

[Alex Kiernan]

On Wed, Nov 22, 2023 at 2:18 PM Yash Shinde <Yash.Shinde@windriver.com> wrote:
>
>
> On 21-11-2023 21:27, Alex Kiernan wrote:
> > CAUTION: This email comes from a non Wind River email account!
> > Do not click links or open attachments unless you recognize the sender and know the content is safe.
> >
> > On Tue, Nov 21, 2023 at 11:30 AM Yash Shinde <Yash.Shinde@windriver.com> wrote:
> >>
> >> On 20-11-2023 21:57, Alex Kiernan wrote:
> >>> CAUTION: This email comes from a non Wind River email account!
> >>> Do not click links or open attachments unless you recognize the sender and know the content is safe.
> >>>
> >>> I've finally got this running in a way I can usefully inspect and try
> >>> things and it's beginning to feel like we've headed into something
> >>> which simply isn't supported.
> >>>
> >>> The failure is this:
> >>>
> >>>     thread 'main' panicked at src/tools/compiletest/src/common.rs:519:30:
> >>>     no entry found for key
> >>>
> >>> Which is empty because the mapping list is collected by running rustc
> >>> with -Z unstable-options --print=all-target-specs-json, but that fails
> >>> because we're not running a nightly compiler:
> >>>
> >>>     error: the option `Z` is only accepted on the nightly compiler
> >>>     help: consider switching to a nightly toolchain: `rustup default nightly`
> >>>     note: selecting a toolchain with `+toolchain` arguments require a
> >>> rustup proxy; see
> >>> <https://rust-lang.github.io/rustup/concepts/index.html>
> >>>     note: for more information about Rust's stability policy, see
> >>> <https://doc.rust-lang.org/book/appendix-07-nightly-rust.html#unstable-features>
> >>>
> >>> Basically feels like running the test suite is only supported on the
> >>> nightlies...
> >> I saw the same error earlier.
> >>
> >> Investigating further, we understood that the issue is not related to
> >> running on nightly channel. The '-Z unstable options' were also present
> >> in earlier rust versions and removing of the newly added -Z options
> >> after 1.70 also didn't solve the problem.
> >>
> > There's a fair number of changes in the test framework around the code
> > that calls out to rustc with new paths that use
> > --print=all-target-specs-json unconditionally. Assuming I've read it
> > right, -Zunstable-options is only evaluated at the point where you
> > actually try and use such an option which is covered by
> > unstable-options, so just passing it in doesn't get you a failure.
> >
> >> On doing further analysis, we found that the custom target config/specs
> >> were not fetched during the rust build.
> >>
> >> The log for reference is as follows:
> >>
> >> Building stage1 library artifacts (x86_64-unknown-linux-gnu ->
> >> x86_64-poky-linux-gnu)
> >> error: failed to run `rustc` to learn about target-specific information
> >>
> >> Caused by:
> >>     process didn't exit successfully: `
> >> build-st/tmp/work/core2-64-poky-linux/rust/1.72.0/rustc-1.72.0-src/build/bootstrap/debug/rustc
> >> - --crate-name ___ --print=file-names -Csymbol-mangling-version=legacy
> >> -Zunstable-options -Zunstable-options '--check-cfg=values(bootstrap)'
> >> '--check-cfg=values(stdarch_intel_sde)'
> >> '--check-cfg=values(no_fp_fmt_parse)'
> >> '--check-cfg=values(no_global_oom_handling)' '--check-cfg=values(no_rc)'
> >> '--check-cfg=values(no_sync)' '--check-cfg=values(freebsd12)'
> >> '--check-cfg=values(freebsd13)'
> >> '--check-cfg=values(backtrace_in_libstd)'
> >> '--check-cfg=values(target_env,"libnx")'
> >> '--check-cfg=values(target_arch,"asmjs","spirv","nvptx","xtensa")'
> >> -Zmacro-backtrace -Clink-args=-Wl,-z,origin
> >> '-Clink-args=-Wl,-rpath,$ORIGIN/../lib' -Csplit-debuginfo=off
> >> -Cprefer-dynamic -Zinline-mir -Cembed-bitcode=yes
> >> '-Zcrate-attr=doc(html_root_url="https://doc.rust-lang.org/1.72.0/")'
> >> --target x86_64-poky-linux-gnu --crate-type bin --crate-type rlib
> >> --crate-type dylib --crate-type cdylib --crate-type staticlib
> >> --crate-type proc-macro --print=sysroot --print=split-debuginfo
> >> --print=crate-name --print=cfg` (exit status: 1)
> >>     --- stderr
> >>     error: Error loading target specification: Could not find
> >> specification for target "x86_64-poky-linux-gnu". Run `rustc --print
> >> target-list` for a list of built-in targets
> >>
> > Out of interest, did you try 1.71.0 to see if that failed? Based on
> > reading the commit log for src/tools/compiletest/src/common.rs, I
> > suspect it will fail.
> >
> Yes, that fails in 1.71.0 also. The 1.71 build progresses to check some
> test suites before failing as follows-
>

I think this is where our problems start:

commit c075691b825dce55e4261aba570cd2d9665950c0
Author: Pietro Albini <pietro.albini@ferrous-systems.com>
Date:   Thu Mar 9 17:14:25 2023 +0100

    use --print=all-target-specs-json for stage1+

But there's a bunch of commits in 1.71.0 which build on that, but
which aren't isolated to just that code, so I'm struggling to "revert"
just that change to prove it.

If it's useful I brought up all the intermediate versions into my tree
if it's useful (though I've only build tested 1.71.0 and 1.74.0).

-- 
Alex Kiernan