strace: upgrade to 4.9, supports ARC
PATCH 1/1 strace: upgrade to 4.9, supports ARC
Also get rid of the backport patches which are already present upstream.
Tested on ARC with LFS
--- package/strace/Config.in | 2 +- .../strace-02-fix-ptrace_peeksiginfo_args.patch | 100 --------------------- package/strace/strace-03-update-configure.patch | 31 ------- ...xtensa-sort-values-in-struct_user_offsets.patch | 76 ---------------- package/strace/strace.mk | 2 +- 5 files changed, 2 insertions(+), 209 deletions(-) delete mode 100644 package/strace/strace-02-fix-ptrace_peeksiginfo_args.patch delete mode 100644 package/strace/strace-03-update-configure.patch delete mode 100644 package/strace/strace-04-xtensa-sort-values-in-struct_user_offsets.patch
package/strace/Config.in +1 -1
@@ -1,6 +1,6 @@ config BR2_PACKAGE_STRACE bool "strace" - depends on !BR2_arc && !BR2_nios2 + depends on !BR2_nios2 help A useful diagnostic, instructional, and debugging tool. Allows you to track what system calls a program makes
package/strace/strace-02-fix-ptrace_peeksiginfo_args.patch deleted +0 -100
100 lines not on this page · raw message
package/strace/strace-03-update-configure.patch deleted +0 -31
@@ -1,31 +0,0 @@ ---- a/configure 2013-12-02 12:31:53.257634249 +1100 -+++ b/configure 2013-12-02 12:32:17.726085888 +1100 -@@ -6389,6 +6389,16 @@ - - - fi -+ac_fn_c_check_type "$LINENO" "struct ptrace_peeksiginfo_args" "ac_cv_type_struct_ptrace_peeksiginfo_args" "#include <sys/ptrace.h> -+" -+if test "x$ac_cv_type_struct_ptrace_peeksiginfo_args" = xyes; then : -+ -+cat >>confdefs.h <<_ACEOF -+#define HAVE_STRUCT_PTRACE_PEEKSIGINFO_ARGS 1 -+_ACEOF -+ -+ -+fi - - - ac_fn_c_check_type "$LINENO" "struct user_desc" "ac_cv_type_struct_user_desc" "#include <asm/ldt.h> ---- a/config.h.in.orig 2013-12-02 12:44:56.795927681 +1100 -+++ b/config.h.in 2013-12-02 12:45:07.524122894 +1100 -@@ -253,6 +253,9 @@ - /* Define to 1 if `msg_control' is a member of `struct msghdr'. */ - #undef HAVE_STRUCT_MSGHDR_MSG_CONTROL - -+/* Define to 1 if the system has the type `struct ptrace_peeksiginfo_args'. */ -+#undef HAVE_STRUCT_PTRACE_PEEKSIGINFO_ARGS -+ - /* Define to 1 if the system has the type `struct pt_all_user_regs'. */ - #undef HAVE_STRUCT_PT_ALL_USER_REGS -
package/strace/strace-04-xtensa-sort-values-in-struct_user_offsets.patch deleted +0 -76
76 lines not on this page · raw message
package/strace/strace.mk +1 -1
@@ -4,7 +4,7 @@ # ################################################################################ -STRACE_VERSION = 4.8 +STRACE_VERSION = 4.9 STRACE_SOURCE = strace-$(STRACE_VERSION).tar.xz STRACE_SITE = http://downloads.sourceforge.net/project/strace/strace/$(STRACE_VERSION) STRACE_LICENSE = BSD-3c
Signature · 1 line
1.8.3.2
PATCH 1/1 strace: upgrade to 4.9, supports ARC
Quoted 4 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
"Vineet" == Vineet Gupta <[email protected]> writes:
Also get rid of the backport patches which are already present upstream. Tested on ARC with LFS
It indeed seems to have problems with !LFS (on ARM atleast):
file.c: In function ?sys_getdents64?:
file.c:2155:6: error: dereferencing pointer to incomplete type file.c:2156:6: error: dereferencing pointer to incomplete type file.c:2158:28: error: dereferencing pointer to incomplete type file.c:2161:6: error: dereferencing pointer to incomplete type file.c:2161:19: error: dereferencing pointer to incomplete type file.c:2163:9: error: dereferencing pointer to incomplete type file.c:2167:9: error: dereferencing pointer to incomplete type
Do you know if this is fixable, or should we just make it depend on BR2_LARGEFILE?
Signature · 1 line
Bye, Peter Korsgaard
PATCH 1/1 strace: upgrade to 4.9, supports ARC
Hi Peter,
+CC Noam too who has faced strace !LFS issues /
Quoted 15 lines · Peter Korsgaard
Quoted 3 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
"Vineet" == Vineet Gupta <[email protected]> writes:
Also get rid of the backport patches which are already present upstream. Tested on ARC with LFS
It indeed seems to have problems with !LFS (on ARM atleast):
file.c: In function ?sys_getdents64?:
file.c:2155:6: error: dereferencing pointer to incomplete type file.c:2156:6: error: dereferencing pointer to incomplete type file.c:2158:28: error: dereferencing pointer to incomplete type file.c:2161:6: error: dereferencing pointer to incomplete type file.c:2161:19: error: dereferencing pointer to incomplete type file.c:2163:9: error: dereferencing pointer to incomplete type file.c:2167:9: error: dereferencing pointer to incomplete type
Quoted 2 lines
Do you know if this is fixable, or should we just make it depend on
BR2_LARGEFILE?
It seems patch-01 (which I didn't remove) would already handle the !LFS case, but that apparently is not case as u show above.
Can u please check if the 4.8 strace builds OK w/o !LFS. If so that would mean something in upstream strace renders the patch-01 as useless or needing modification.
Otherwise the quick fix is to just make strace depend on LFS.
Thx,
-Vineet
PATCH 1/1 strace: upgrade to 4.9, supports ARC
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
"Vineet" == Vineet Gupta <[email protected]> writes:
Hi,
Quoted 8 lines
Quoted 2 lines
Do you know if this is fixable, or should we just make it depend on
BR2_LARGEFILE?It seems patch-01 (which I didn't remove) would already handle the !LFS case, but that apparently is not case as u show above.
Can u please check if the 4.8 strace builds OK w/o !LFS. If so that would mean something in upstream strace renders the patch-01 as useless or needing modification.
Yes, strace 4.8 builds ok without LFS.
So either we extend patch 01 or we drop the patch and make strace depend on LFS (which is nice as we get rid of a patch, but less nice as it is then more difficult to debug !LFS setups).
Anybody feel strongly for one or the other?
Signature · 1 line
Bye, Peter Korsgaard
PATCH 1/1 strace: upgrade to 4.9, supports ARC
Quoted 14 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
"Peter" == Peter Korsgaard <[email protected]> writes:
"Vineet" == Vineet Gupta <[email protected]> writes:
Hi,
Quoted 8 lines
Quoted 2 lines
Do you know if this is fixable, or should we just make it depend on
BR2_LARGEFILE?It seems patch-01 (which I didn't remove) would already handle the !LFS case, but that apparently is not case as u show above.
Can u please check if the 4.8 strace builds OK w/o !LFS. If so that would mean something in upstream strace renders the patch-01 as useless or needing modification.
Quoted 7 lines
Yes, strace 4.8 builds ok without LFS.
So either we extend patch 01 or we drop the patch and make strace depend on LFS (which is nice as we get rid of a patch, but less nice as it is then more difficult to debug !LFS setups).
Anybody feel strongly for one or the other?
I didn't hear any feedback, so I decided to drop the patch and make strace depend on LFS.
Committed, thanks.
Signature · 1 line
Bye, Peter Korsgaard