| Message ID | 20180712221730.GK14131@decadent.org.uk (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
2018-07-13 7:17 GMT+09:00 Ben Hutchings <ben@decadent.org.uk>: > In the $(run-cmd) macro, add a trap on EXIT that prints the > full command line. Remove the trap after running the command(s) > successfully. > > (A more straightforward approach would be to use "if" or "||" to test > for failure, but that doesn't work. Some command lines given to > $(run-cmd) have multiple commands separated by semi-colons, and the > caller must run "set -e" to enable exit-on-error. Testing the result > of such a command line, even if it is probably grouped and run in a > sub-shell, inhibits exit-on-error and would cause some errors to be > ignored.) > > Signed-off-by: Ben Hutchings <ben@decadent.org.uk> OK, this seems useful to me. > --- > scripts/Kbuild.include | 5 ++++- > 1 file changed, 4 insertions(+), 1 deletion(-) > > diff --git a/scripts/Kbuild.include b/scripts/Kbuild.include > index 8778ae4a3476..c2525aaa36ac 100644 > --- a/scripts/Kbuild.include > +++ b/scripts/Kbuild.include > @@ -224,7 +224,10 @@ flags = $(foreach o,$($(1)),$(if $(filter -I%,$(o)),$(call addtree,$(o)),$(o))) > echo-cmd = $(if $($(quiet)cmd_$(1)),\ > echo ' $(call escsq,$($(quiet)cmd_$(1)))$(echo-why)';) > > -run-cmd = $(echo-cmd) $(cmd_$(1)) > +trap-cmd-failed = trap 'test $$? = 0 || echo Failed command: '\''$(call escsq,$(call escsq,$(cmd_$(1))))'\' EXIT > +untrap-cmd-failed = trap - EXIT Why $(call escsq,...) twice? I think ... echo Failed command: '\''$(call escsq,$(cmd_$(1)))'\' EXIT is enough. I see another problem for this. Make cannot detect the failure of $(call cmd,...) --------------(sample code)--------------- extra-y := foo.txt quiet_cmd_always_fail = FAIL $@ cmd_always_fail = /bin/false $(obj)/foo.txt: $(call cmd,always_fail) ------------(sample code end)------------- This should terminate the build. With your series, this always succeeds because it returns the exit code of 'trap - EXIT', which is 0. Maybe, $(cmd_$(1)) && $(untrap-cmd-failed) ? > + > +run-cmd = $(echo-cmd) $(if $(cmd_$(1)), $(if $(quiet), $(trap-cmd-failed); $(cmd_$(1)); $(untrap-cmd-failed), $(cmd_$(1)))) > > # printing commands > cmd = @$(run-cmd)
diff --git a/scripts/Kbuild.include b/scripts/Kbuild.include index 8778ae4a3476..c2525aaa36ac 100644 --- a/scripts/Kbuild.include +++ b/scripts/Kbuild.include @@ -224,7 +224,10 @@ flags = $(foreach o,$($(1)),$(if $(filter -I%,$(o)),$(call addtree,$(o)),$(o))) echo-cmd = $(if $($(quiet)cmd_$(1)),\ echo ' $(call escsq,$($(quiet)cmd_$(1)))$(echo-why)';) -run-cmd = $(echo-cmd) $(cmd_$(1)) +trap-cmd-failed = trap 'test $$? = 0 || echo Failed command: '\''$(call escsq,$(call escsq,$(cmd_$(1))))'\' EXIT +untrap-cmd-failed = trap - EXIT + +run-cmd = $(echo-cmd) $(if $(cmd_$(1)), $(if $(quiet), $(trap-cmd-failed); $(cmd_$(1)); $(untrap-cmd-failed), $(cmd_$(1)))) # printing commands cmd = @$(run-cmd)
In the $(run-cmd) macro, add a trap on EXIT that prints the full command line. Remove the trap after running the command(s) successfully. (A more straightforward approach would be to use "if" or "||" to test for failure, but that doesn't work. Some command lines given to $(run-cmd) have multiple commands separated by semi-colons, and the caller must run "set -e" to enable exit-on-error. Testing the result of such a command line, even if it is probably grouped and run in a sub-shell, inhibits exit-on-error and would cause some errors to be ignored.) Signed-off-by: Ben Hutchings <ben@decadent.org.uk> --- scripts/Kbuild.include | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)