| Message ID | 20170409205128.11560-5-arnout@mind.be |
|---|---|
| State | Superseded |
| Headers | show |
Hello, On Sun, 9 Apr 2017 22:51:25 +0200, Arnout Vandecappelle (Essensium/Mind) wrote: > Some lines from the toolchain config may be removed due to dependency > issues. Currently this is covered by explicit conditions in the > autobuild-run script, e.g. checking that libc is not glibc before > enabling BR2_STATIC_LIBS. However, that binds this script pretty > tightly to the logic in Buildroot itself. Seems like a good idea! > + # Check that the toolchain configuration is still present > + # Report all the missing ones > + toolchaincomplete = True > + for toolchainline in kwargs['config']: > + if toolchainline not in configlines: > + log_write(log, "WARN: missing toolchain config line: %s" % toolchainline[:-1]) We should remove this warning, it's not a warning at all. It is a perfectly normal situation that a configuration gets rejected here, for example when glibc+static is used. So having a warning everytime a completely regular situation occurs doesn't seem like a good idea. Thanks! Thomas
On 10-04-17 10:28, Thomas Petazzoni wrote: > Hello, > > On Sun, 9 Apr 2017 22:51:25 +0200, Arnout Vandecappelle > (Essensium/Mind) wrote: >> Some lines from the toolchain config may be removed due to dependency >> issues. Currently this is covered by explicit conditions in the >> autobuild-run script, e.g. checking that libc is not glibc before >> enabling BR2_STATIC_LIBS. However, that binds this script pretty >> tightly to the logic in Buildroot itself. > > Seems like a good idea! > >> + # Check that the toolchain configuration is still present >> + # Report all the missing ones >> + toolchaincomplete = True >> + for toolchainline in kwargs['config']: >> + if toolchainline not in configlines: >> + log_write(log, "WARN: missing toolchain config line: %s" % toolchainline[:-1]) > > We should remove this warning, it's not a warning at all. It is > a perfectly normal situation that a configuration gets rejected here, > for example when glibc+static is used. So having a warning everytime a > completely regular situation occurs doesn't seem like a good idea. Yes of course. I needed that for debugging but it indeed shouldn't be there in production. Regards, Arnout
diff --git a/scripts/autobuild-run b/scripts/autobuild-run index 1df1ce0..50808b6 100755 --- a/scripts/autobuild-run +++ b/scripts/autobuild-run @@ -366,6 +366,16 @@ def is_toolchain_usable(**kwargs): with open(os.path.join(outputdir, ".config")) as configf: configlines = configf.readlines() + # Check that the toolchain configuration is still present + # Report all the missing ones + toolchaincomplete = True + for toolchainline in kwargs['config']: + if toolchainline not in configlines: + log_write(log, "WARN: missing toolchain config line: %s" % toolchainline[:-1]) + toolchaincomplete = False + if not toolchaincomplete: + return False + # The latest Linaro toolchains on x86-64 hosts requires glibc # 2.14+ on the host. if platform.machine() == 'x86_64': @@ -564,7 +574,7 @@ def gen_config(**kwargs): log_write(log, "ERROR: cannot oldconfig") return -1 - if not is_toolchain_usable(**kwargs): + if not is_toolchain_usable(config=config["contents"], **kwargs): return -1 # Now, generate the random selection of packages, and fixup
Some lines from the toolchain config may be removed due to dependency issues. Currently this is covered by explicit conditions in the autobuild-run script, e.g. checking that libc is not glibc before enabling BR2_STATIC_LIBS. However, that binds this script pretty tightly to the logic in Buildroot itself. So instead, just check that the toolchain configuration is still valid after running 'olddefconfig', and discard the configuration if it isn't. This is similar to how it's done in the test-pkg script. We also report all the missing lines in the log file. Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be> --- scripts/autobuild-run | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-)