firefox 56 emerge failure - backtrace-sys rusty cargo.

[eccerr0r]

I think this might be a one-off alas wonder if anyone has seen this before:

Code: Select all

error: the listed checksum of `/tmp/portage/www-client/firefox-56.0-r1/work/firefox-56.0/third_party/rust/backtrace-sys/src/libbacktrace/configure` has changed:
expected: 2b45c89d06809c6ed49d70bfdf51a93e44370fc81f689169a073f47b14b366e5
actual:   937feb5d999f32d2917b8f105428ee11add2729c2318a27fb3509e8f4ba7d06b

directory sources are not intended to be edited, if modifications are required then it is recommended that [replace] is used with a forked copy of the source
make[4]: *** [/tmp/portage/www-client/firefox-56.0-r1/work/firefox-56.0/config/rules.mk:1020: force-cargo-program-build] Error 101

It seems that I somehow downloaded an incorrect version though portage (???) seems to say it's OK?

What's going on here?

EDIT: OOPS. I should have reverted back to the original ebuild that's on the portage mirrors before reporting this, though the hack to my ebuild in my local repository should have been benign to this... That's why nobody sees an -r1 :)

[duby2291]

I'm not sure if this is related, maybe not, but perhaps it is. I had to do this "echo >=dev-util/cargo-0.21.0-r1 >> /etc/portage/package.mask && emerge -1 =dev-util/cargo-0.21.0"

I don't understand why, but the build failure I experienced indicated that supported cargo versions are hardcoded. So the only solution is to make sure the version of cargo that firefox expects is installed. You should use the unmodified stock firefox sources with the specific version of cargo that it is coded to expect.

[eccerr0r]

I have cargo-0.21.0 installed, I think this requirement was in the ebuild. Not sure if rust/backtrace-sys somehow is wrong, but it should have been the same.

The patch was to try to include geckodriver and I saw that the ebuild was doing something similar, so the first pass I added it there. This generated that weird error which makes no sense as I wasn't editing any rust files.

So I did another experiment: I changed my plan from a hacked ebuild to an epatch file, which I probably should have done in the first place, if it were clear how to do it (I'm no Mozilla developer.)

Still failed.

I think the firefox-source package itself is busted. Along with the rant about the rust packaging, this pisses me off...

[cjmayo]

I don't think it is the source but because the ebuild recreates a couple of files and therefore they have a different checksum.

I've worked around this by adding to the end of src_prepare():

Code: Select all

	sed -i "s:ee4019f80b993cdb1bd13bf39434ad4f1ef15ae57a97594c53fb7b1d76d776a4:$(sha256sum ${S}/third_party/rust/backtrace-sys/src/libbacktrace/ltmain.sh | cut -d ' ' -f 1):" \
		"${S}"/third_party/rust/backtrace-sys/.cargo-checksum.json || die

	sed -i "s:2b45c89d06809c6ed49d70bfdf51a93e44370fc81f689169a073f47b14b366e5:$(sha256sum ${S}/third_party/rust/backtrace-sys/src/libbacktrace/configure | cut -d ' ' -f 1):" \
		"${S}"/third_party/rust/backtrace-sys/.cargo-checksum.json || die

But then it fails with:

Code: Select all

process didn't exit successfully: `/var/tmp/portage/www-client/firefox-57.0.1-r1/work/firefox-57.0.1/ff/testing/geckodriver/./release/build/bzip2-sys-2df4d9d26581b387/build-script-build` (exit code: 101)
--- stdout
TARGET = Some("x86_64-unknown-linux-gnu")
OPT_LEVEL = Some("3")
TARGET = Some("x86_64-unknown-linux-gnu")
HOST = Some("x86_64-unknown-linux-gnu")
TARGET = Some("x86_64-unknown-linux-gnu")
TARGET = Some("x86_64-unknown-linux-gnu")
HOST = Some("x86_64-unknown-linux-gnu")
CC_x86_64-unknown-linux-gnu = None
CC_x86_64_unknown_linux_gnu = None
HOST_CC = Some("/usr/bin/x86_64-pc-linux-gnu-gcc -std=gnu99")
TARGET = Some("x86_64-unknown-linux-gnu")
HOST = Some("x86_64-unknown-linux-gnu")
CFLAGS_x86_64-unknown-linux-gnu = None
CFLAGS_x86_64_unknown_linux_gnu = None
HOST_CFLAGS = None
CFLAGS = Some(" -Wall -Wempty-body -Wignored-qualifiers -Wpointer-arith -Wsign-compare -Wtype-limits -Wunreachable-code -Wno-error=maybe-uninitialized -Wno-error=deprecated-declarations -Wno-error=array-bounds -Wno-error=free-nonheap-object -Wformat -Wformat-security -march=ivybridge -pipe -fno-strict-aliasing -ffunction-sections -fdata-sections -fno-math-errno -pthread -pipe  -freorder-blocks -O2 -fomit-frame-pointer ")
PROFILE = Some("release")
running: "/usr/bin/x86_64-pc-linux-gnu-gcc -std=gnu99" "-O3" "-ffunction-sections" "-fdata-sections" "-fPIC" "-Wall" "-Wempty-body" "-Wignored-qualifiers" "-Wpointer-arith" "-Wsign-compare" "-Wtype-limits" "-Wunreachable-code" "-Wno-error=maybe-uninitialized" "-Wno-error=deprecated-declarations" "-Wno-error=array-bounds" "-Wno-error=free-nonheap-object" "-Wformat" "-Wformat-security" "-march=ivybridge" "-pipe" "-fno-strict-aliasing" "-ffunction-sections" "-fdata-sections" "-fno-math-errno" "-pthread" "-pipe" "-freorder-blocks" "-O2" "-fomit-frame-pointer" "-m64" "-I" "bzip2-1.0.6" "-DBZ_NO_STDIO" "-o" "/var/tmp/portage/www-client/firefox-57.0.1-r1/work/firefox-57.0.1/ff/testing/geckodriver/./x86_64-unknown-linux-gnu/release/build/bzip2-sys-0ba348e4bd239980/out/bzip2-1.0.6/blocksort.o" "-c" "bzip2-1.0.6/blocksort.c"


failed to execute command: No such file or directory (os error 2)
Is `x86_64-pc-linux-gnu-gcc -std=gnu99` not installed?

Looks like the problem is that the value of HOST_CC includes an option but then when it is run the first string needs to be just the executable. Don't have a fix for this.

(Suggest adding at least geckodriver to the post title).

[eccerr0r]

Indeed I was trying to get geckodriver to build. I ended up finding https://github.com/gentoo/gentoo-rust which contains a working geckodriver ebuild. But thanks for the insight into hacking the needed files to update the checksums.