audit updates and fixes
Audit fixes and missing bits; compared to the last attempt there are two added fixes from Amy... Please, pull:git://git.kernel.org/pub/scm/linux/kernel/git/viro/audit-current.git/ audit.b28
Shortlog:
Al Viro:
syscall classes hookup for ppc and s390
audit: more syscall classes added
audit: AUDIT_PERM support
Amy Griffis:
sanity check audit_buffer
update audit rule change messages
Steve Grubb:
fix ppid bug in 2.6.18 kernel
Diffstat:
arch/i386/kernel/audit.c | 28 +++++++++++++++ arch/ia64/ia32/audit.c | 26 ++++++++++++++ arch/ia64/kernel/audit.c | 35 +++++++++++++++++++ arch/powerpc/kernel/Makefile | 2 + arch/powerpc/kernel/audit.c | 66 +++++++++++++++++++++++++++++++++++++ arch/powerpc/kernel/compat_audit.c | 38 +++++++++++++++++++++ arch/s390/kernel/Makefile | 4 +- arch/s390/kernel/audit.c | 66 +++++++++++++++++++++++++++++++++++++ arch/s390/kernel/compat_audit.c | 38 +++++++++++++++++++++ arch/x86_64/ia32/audit.c | 26 ++++++++++++++ arch/x86_64/kernel/audit.c | 35 +++++++++++++++++++ include/asm-generic/audit_read.h | 8 ++++ include/asm-generic/audit_write.h | 11 ++++++ include/linux/audit.h | 11 ++++++ kernel/audit.c | 6 +++ kernel/audit.h | 1 kernel/auditfilter.c | 37 +++++++++++++++++--- kernel/auditsc.c | 51 ++++++++++++++++++++++++++++ 18 files changed, 483 insertions(+), 6 deletions(-)
Quoted 3 lines · On Mon, 11 Sep 2006 18:37:49 +0100 Al Viro <[email protected]> wrote:
Audit fixes and missing bits; compared to the last attempt there are two added fixes from Amy... Please, pull:
git://git.kernel.org/pub/scm/linux/kernel/git/viro/audit-current.git/audit.b28
This seems a bit.... late.
The patch adds a whole bunch of new extern-decls-in-C. It'd be nice to get those into a header file sometime.
And sparc64 allmodconfig broke.
kernel/built-in.o(.text+0x36834): In function `audit_filter_rules':
: undefined reference to `audit_classify_syscall'
make: *** [.tmp_vmlinux1] Error 1
Possibly because the audit git tree fell out of -mm ages ago, partly because its owner (ahem) ignored my request to resync it with upstream. I have new scripts from Junio and shall try again.
Duh... Two trivial fixes:
* make sure that all non-biarch targets get hookup.
* deal with sparc64 correctly.
[ObGit: git-diff after git-mv + modify a couple of lines generates deletion+new file instead of rename. Weird...]
Please, pull from git.kernel.org/pub/scm/linux/kernel/git/viro/audit-current.git/ audit.b29
Al Viro:
syscall class hookup for all normal targets
sparc64 audit syscall classes hookup
arch/i386/kernel/audit.c | 51 --------------------------- b/arch/i386/kernel/Makefile | 1 b/arch/ia64/Kconfig | 4 ++ b/arch/powerpc/Kconfig | 4 ++ b/arch/s390/Kconfig | 4 ++ b/arch/sparc64/Kconfig | 4 ++ b/arch/sparc64/kernel/Makefile | 3 + b/arch/sparc64/kernel/audit.c | 66 +++++++++++++++++++++++++++++++++++ b/arch/sparc64/kernel/compat_audit.c | 37 +++++++++++++++++++ b/arch/x86_64/Kconfig | 4 ++ b/lib/Kconfig | 5 ++ b/lib/Makefile | 1 b/lib/audit.c | 53 ++++++++++++++++++++++++++++ 13 files changed, 185 insertions(+), 52 deletions(-)
Quoted 3 lines · Al Viro
[ObGit: git-diff after git-mv + modify a couple of lines generates deletion+new file instead of rename. Weird...]
No it doesn't.
Are you sure you used the "-M" flag to ask for renames?
The other alternative is that the file was so small that the change is considered significant, and it becomes a rewrite.
Linus
Quoted 15 lines · Al Viro
arch/i386/kernel/audit.c | 51 --------------------------- b/arch/i386/kernel/Makefile | 1 b/arch/ia64/Kconfig | 4 ++ b/arch/powerpc/Kconfig | 4 ++ b/arch/s390/Kconfig | 4 ++ b/arch/sparc64/Kconfig | 4 ++ b/arch/sparc64/kernel/Makefile | 3 + b/arch/sparc64/kernel/audit.c | 66 +++++++++++++++++++++++++++++++++++ b/arch/sparc64/kernel/compat_audit.c | 37 +++++++++++++++++++ b/arch/x86_64/Kconfig | 4 ++ b/lib/Kconfig | 5 ++ b/lib/Makefile | 1 b/lib/audit.c | 53 ++++++++++++++++++++++++++++ 13 files changed, 185 insertions(+), 52 deletions(-)
Btw, if you do "git diff --stat --summary -M", you'd have gotten
arch/i386/kernel/Makefile | 1 -
arch/ia64/Kconfig | 4 ++
arch/powerpc/Kconfig | 4 ++
arch/s390/Kconfig | 4 ++
arch/sparc64/Kconfig | 4 ++
arch/sparc64/kernel/Makefile | 3 ++
arch/sparc64/kernel/audit.c | 66 ++++++++++++++++++++++++++++++++++++
arch/sparc64/kernel/compat_audit.c | 37 ++++++++++++++++++++
arch/x86_64/Kconfig | 4 ++
lib/Kconfig | 5 +++
lib/Makefile | 1 +
{arch/i386/kernel => lib}/audit.c | 2 +
12 files changed, 134 insertions(+), 1 deletions(-)
create mode 100644 arch/sparc64/kernel/audit.c
create mode 100644 arch/sparc64/kernel/compat_audit.c
rename arch/i386/kernel/audit.c => lib/audit.c (97%)instead, which is much nicer. It shows the rename, and the summary info also points out that some of the files were newly created.
Linus
Quoted 34 lines · Linus Torvalds
Quoted 6 lines
arch/i386/kernel/audit.c | 51 --------------------------- b/arch/i386/kernel/Makefile | 1 b/arch/ia64/Kconfig | 4 ++ b/arch/powerpc/Kconfig | 4 ++ b/arch/s390/Kconfig | 4 ++ b/arch/sparc64/Kconfig | 4 ++Quoted 8 lines
b/arch/sparc64/kernel/Makefile | 3 + b/arch/sparc64/kernel/audit.c | 66 +++++++++++++++++++++++++++++++++++ b/arch/sparc64/kernel/compat_audit.c | 37 +++++++++++++++++++ b/arch/x86_64/Kconfig | 4 ++ b/lib/Kconfig | 5 ++ b/lib/Makefile | 1 b/lib/audit.c | 53 ++++++++++++++++++++++++++++ 13 files changed, 185 insertions(+), 52 deletions(-)Btw, if you do "git diff --stat --summary -M", you'd have gotten
arch/i386/kernel/Makefile | 1 - arch/ia64/Kconfig | 4 ++ arch/powerpc/Kconfig | 4 ++ arch/s390/Kconfig | 4 ++ arch/sparc64/Kconfig | 4 ++ arch/sparc64/kernel/Makefile | 3 ++ arch/sparc64/kernel/audit.c | 66 ++++++++++++++++++++++++++++++++++++ arch/sparc64/kernel/compat_audit.c | 37 ++++++++++++++++++++ arch/x86_64/Kconfig | 4 ++ lib/Kconfig | 5 +++ lib/Makefile | 1 + {arch/i386/kernel => lib}/audit.c | 2 + 12 files changed, 134 insertions(+), 1 deletions(-) create mode 100644 arch/sparc64/kernel/audit.c create mode 100644 arch/sparc64/kernel/compat_audit.c rename arch/i386/kernel/audit.c => lib/audit.c (97%)
Quoted 2 lines
instead, which is much nicer. It shows the rename, and the summary info also points out that some of the files were newly created.
I see... FWIW, I'd been using git diff fed through diffstat(1) for that part. Thanks, that's indeed much better.
ObMIPS: how many ABI do we have there, anyway? 4 + IRIX compat? More?