running LTP on uClinux
I was building LTP to test a uClinux system, but encountered some problems. As far as I could tell UCLINUX=1 is not being passed to all testcases when defined at the top level before running make.
export UCLINUX=1
make
I modified the following file and the build proceeded as expected:
mk/testcases.mk 2010-07-01 17:38:10.240628708 -0500 renamed from mk/testcases.mk.orig 2010-04-01 01:23:08.000000000 -0500 +4 -0
@@ -53,3 +53,8 @@ $(APICMDS_DIR) $(LIBLTP_DIR) $(abs_top_builddir)/$(TKI_DIR): %: mkdir -p "$@" + +ifeq ($(UCLINUX),1) +CFLAGS += -D__UCLIBC__ -DUCLINUX +endif
Has anyone else had to make similar changes in order to build LTP for uClinux?
d.marlin
------------------------------------------------------------------------------
This SF.net email is sponsored by
Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list
Quoted 23 lines · David Marlin
I was building LTP to test a uClinux system, but encountered some problems. As far as I could tell UCLINUX=1 is not being passed to all testcases when defined at the top level before running make.
export UCLINUX=1
makeI modified the following file and the build proceeded as expected:
mk/testcases.mk 2010-07-01 17:38:10.240628708 -0500 renamed from mk/testcases.mk.orig 2010-04-01 01:23:08.000000000 -0500 +0
-0@@ -53,3 +53,8 @@$(APICMDS_DIR) $(LIBLTP_DIR) $(abs_top_builddir)/$(TKI_DIR): %:
mkdir -p "$@"mk/testcases.mk 2010-07-01 17:38:10.240628708 -0500 renamed from mk/testcases.mk.orig 2010-04-01 01:23:08.000000000 -0500 +4
-0+ +ifeq ($(UCLINUX),1) +CFLAGS += -D__UCLIBC__ -DUCLINUX +endif
Quoted 2 lines
Has anyone else had to make similar changes in order to build LTP for uClinux?
It should be CPPFLAGS, and the definition should be in env_post.mk; I never had a UCLINUX infrastructure to test with, so this might have fallen between the cracks by accident about 8 months ago.
Let me know how this patch works for you.
Thanks,
-Garrett
include/mk/env_post.mk +4 -0
@@ -37,6 +37,10 @@ CPPFLAGS += -I$(top_srcdir)/include -I$(top_builddir)/include LDFLAGS += -L$(top_builddir)/lib +ifeq ($(UCLINUX),1) +CPPFLAGS += -D__UCLIBC__ -DUCLINUX +endif + MAKE_TARGETS ?= $(notdir $(patsubst %.c,%,$(wildcard $(abs_srcdir)/*.c))) MAKE_TARGETS := $(filter-out $(FILTER_OUT_MAKE_TARGETS),$(MAKE_TARGETS))
------------------------------------------------------------------------------
This SF.net email is sponsored by
Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list
Quoted 3 lines · Garrett Cooper
It should be CPPFLAGS, and the definition should be in env_post.mk; I never had a UCLINUX infrastructure to test with, so this might have fallen between the cracks by accident about 8 months ago.
building for a nommu target should work even when you're running a mmu toolchain for an mmu target. the testcases simply assume things like fork() wont work and so fall back to ones that work in both places (like vfork). the tests should be able to run & pass even with a mmu.
Quoted 1 line
Let me know how this patch works for you.
the logic in lib/Makefile should be unified with this -mike
------------------------------------------------------------------------------
This SF.net email is sponsored by
Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list
Quoted 9 lines · Mike Frysinger
Quoted 3 lines · Garrett Cooper
It should be CPPFLAGS, and the definition should be in env_post.mk; I never had a UCLINUX infrastructure to test with, so this might have fallen between the cracks by accident about 8 months ago.
building for a nommu target should work even when you're running a mmu toolchain for an mmu target. the testcases simply assume things like
fork()wont work and so fall back to ones that work in both places (like vfork). the tests should be able to run & pass even with a mmu.
Ok.
Quoted 3 lines
Quoted 1 line
Let me know how this patch works for you.
the logic in lib/Makefile should be unified with this
lib.mk picks up env_post.mk, so it's already caught there. Or are you referring to how libltp is built and what headers are picked up from include/ ?
Thanks!
-Garrett
------------------------------------------------------------------------------
This SF.net email is sponsored by
Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list
Quoted 9 lines · Garrett Cooper
Quoted 4 lines · Mike Frysinger
Quoted 1 line · Garrett Cooper
Let me know how this patch works for you.
the logic in lib/Makefile should be unified with this
lib.mk picks up env_post.mk, so it's already caught there. Or are you referring to how libltp is built and what headers are picked up from
include/?
i mean there should not be more than one place in the whole tree which adds -DUCLINUX/etc... if we merge your patch, there will be. i'm not saying your patch is wrong, just that something needs fixing to avoid the aforementioned scenario.
-mike
------------------------------------------------------------------------------
This SF.net email is sponsored by
Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list
Quoted 7 lines · Mike Frysinger
Quoted 6 lines · Garrett Cooper
Quoted 4 lines · Mike Frysinger
Quoted 1 line · Garrett Cooper
Let me know how this patch works for you.
the logic in lib/Makefile should be unified with this
Quoted 8 lines
Quoted 3 lines
lib.mk picks up env_post.mk, so it's already caught there. Or are you referring to how libltp is built and what headers are picked up from
include/?i mean there should not be more than one place in the whole tree which adds -DUCLINUX/etc... if we merge your patch, there will be. i'm not saying your patch is wrong, just that something needs fixing to avoid the aforementioned scenario.
Sure. I had forgotten whether or not there was still a reference to UCLINUX in lib/Makefile. Thanks for the reminder :). -Garrett
------------------------------------------------------------------------------
This SF.net email is sponsored by
Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list
Quoted 23 lines · Garrett Cooper
Quoted 22 lines · David Marlin
I was building LTP to test a uClinux system, but encountered some problems. As far as I could tell UCLINUX=1 is not being passed to all testcases when defined at the top level before running make.
export UCLINUX=1
makeI modified the following file and the build proceeded as expected:
mk/testcases.mk 2010-07-01 17:38:10.240628708 -0500 renamed from mk/testcases.mk.orig 2010-04-01 01:23:08.000000000 -0500 +4
-0@@ -53,3 +53,8 @@ $(APICMDS_DIR) $(LIBLTP_DIR) $(abs_top_builddir)/$(TKI_DIR): %: mkdir -p "$@" + +ifeq ($(UCLINUX),1) +CFLAGS += -D__UCLIBC__ -DUCLINUX +endif
Quoted 8 lines
Quoted 2 lines
Has anyone else had to make similar changes in order to build LTP for uClinux?
It should be CPPFLAGS, and the definition should be in env_post.mk; I never had a UCLINUX infrastructure to test with, so this might have fallen between the cracks by accident about 8 months ago.
Let me know how this patch works for you.
The patch works for me. Thank you.
I am now seeing some other UCLINUX related problems/questions with individual testcases, but will follow up in separate emails on those.
Thanks again,
d.marlin
==========
------------------------------------------------------------------------------
This SF.net email is sponsored by
Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list