Summary: | sys-block/btrace does not respect LDFLAGS | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Diego Elio Pettenò (RETIRED) <flameeyes> |
Component: | New packages | Assignee: | Robin Johnson <robbat2> |
Status: | RESOLVED FIXED | ||
Severity: | QA | ||
Priority: | High | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | |||
Bug Blocks: | 331933 | ||
Attachments: |
Build log
Patch to btrace-1.0.0.ebuild to respect LDFLAGS Replacement for existing btrace-1.0.0-parallel-make.patch |
Description
Diego Elio Pettenò (RETIRED)
2010-09-03 01:05:04 UTC
Created attachment 245811 [details]
Build log
Created attachment 245822 [details, diff]
Patch to btrace-1.0.0.ebuild to respect LDFLAGS
This patch adds a sed expression to insert $(LDFLAGS) in the lines that link binaries.
Also, while testing, I noticed that a high level of parallelism still causes build failures, even with btrace-1.0.0-parallel-make.patch applied. I see why this is, and will attach a replacement patch that is not affected.
Created attachment 245823 [details, diff]
Replacement for existing btrace-1.0.0-parallel-make.patch
The existing patch, although better than nothing, still allows an unsafe parallelism in the case that Make decides to build btreplay/btreplay and btreplay/btrecord in parallel. When it does so, it spawns two copies of "$(MAKE) -C btreplay", which proceed to step on each other in the subdirectory. The fix I chose here is to make both those targets have no rule of their own, but to have them both depend on a single phony target which does descend into btreplay. This way, make performs one descent and compiles everything in that directory, then goes to process the two targets and finds no further work to do.
Fixed in a different manner as your parallel fixes caused it to try and rebuild the btreplay binaries from the .make phony file. |