Github/littlefs
1
Fork 0
Code Actions Activity

Fixed spurious CI failure caused by multiple writers to .o files

Browse Source 
GCC is a bit frustrating here, it really wants to generate every file in
a single command, which _is_ more efficient if our build system could
leverage this. But -fcallgraph-info is a rather novel flag, so we can't
really rely on it for generally compiling and testing littlefs.

The multi-file output gets in the way when we want an explicitly
separate rule for callgraph-info generation. We can't generate the
callgraph-info without generating the objects files.

This becomes a surprsing issue when parallel building (make -j) is used!
Suddenly we might end up with both the .o and .ci rules writing to .o
files, which creates a really difficult to track down issue of corrupted
.o files.

The temporary solution is to use an order-only prerequisite. This still
ends up building the .o files twice, but it's an acceptable tradeoff for
not requiring the -fcallgraph-info for all builds.
This commit is contained in:
Christopher Haster
2022-03-12 14:11:34 -06:00
parent e4adefd1d7
commit 3b495bab79
1 changed files with 15 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Makefile
+15 -21
View File
@@ -124,16 +124,8 @@ coverage:
./scripts/coverage.py $(BUILDDIR)tests/*.toml.info -s $(COVERAGEFLAGS) ./scripts/coverage.py $(BUILDDIR)tests/*.toml.info -s $(COVERAGEFLAGS)
.PHONY: summary .PHONY: summary
summary: $(OBJ) $(CGI) summary: $(BUILDDIR)lfs.csv
$(strip \ ./scripts/summary.py -Y $^ $(SUMMARYFLAGS)
./scripts/code.py $(OBJ) -q -o - $(CODEFLAGS) \
| ./scripts/data.py $(OBJ) -q -m - -o - $(DATAFLAGS) \
| ./scripts/stack.py $(CGI) -q -m - -o - $(STACKFLAGS) \
| ./scripts/structs.py $(OBJ) -q -m - -o - $(STRUCTFLAGS) \
$(if $(COVERAGE),\
| ./scripts/coverage.py $(BUILDDIR)tests/*.toml.info \
-q -m - -o - $(COVERAGEFLAGS)) \
| ./scripts/summary.py -Y $(SUMMARYFLAGS))
# rules # rules
@@ -147,15 +139,13 @@ $(BUILDDIR)lfs.a: $(OBJ)
$(AR) rcs $@ $^ $(AR) rcs $@ $^
$(BUILDDIR)lfs.csv: $(OBJ) $(CGI) $(BUILDDIR)lfs.csv: $(OBJ) $(CGI)
$(strip \ ./scripts/code.py $(OBJ) -q $(CODEFLAGS) -o $@
./scripts/code.py $(OBJ) -q -o - $(CODEFLAGS) \ ./scripts/data.py $(OBJ) -q -m $@ $(DATAFLAGS) -o $@
| ./scripts/data.py $(OBJ) -q -m - -o - $(DATAFLAGS) \ ./scripts/stack.py $(CGI) -q -m $@ $(STACKFLAGS) -o $@
| ./scripts/stack.py $(CGI) -q -m - -o - $(STACKFLAGS) \ ./scripts/structs.py $(OBJ) -q -m $@ $(STRUCTSFLAGS) -o $@
| ./scripts/structs.py $(OBJ) -q -m - -o - $(STRUCTFLAGS) \ $(if $(COVERAGE),\
$(if $(COVERAGE),\ ./scripts/coverage.py $(BUILDDIR)tests/*.toml.info \
| ./scripts/coverage.py $(BUILDDIR)tests/*.toml.info \ -q -m $@ $(COVERAGEFLAGS) -o $@)
-q -m - -o - $(COVERAGEFLAGS)) \
> $@)
$(BUILDDIR)%.o: %.c $(BUILDDIR)%.o: %.c
$(CC) -c -MMD $(CFLAGS) $< -o $@ $(CC) -c -MMD $(CFLAGS) $< -o $@
@@ -163,8 +153,12 @@ $(BUILDDIR)%.o: %.c
$(BUILDDIR)%.s: %.c $(BUILDDIR)%.s: %.c
$(CC) -S $(CFLAGS) $< -o $@ $(CC) -S $(CFLAGS) $< -o $@
$(BUILDDIR)%.ci: %.c # gcc depends on the output file for intermediate file names, so
$(CC) -c -MMD -fcallgraph-info=su $(CFLAGS) $< -o $(@:.ci=.o) # we can't omit to .o output. We also need to serialize with the
# normal .o rule because otherwise we can end up with multiprocess
# problems with two instances of gcc modifying the same .o
$(BUILDDIR)%.ci: %.c | $(BUILDDIR)%.o
$(CC) -c -MMD -fcallgraph-info=su $(CFLAGS) $< -o $|
# clean everything # clean everything
.PHONY: clean .PHONY: clean