Use generic bug.
From: Heiko Carstens <[email protected]>
[S390] Use generic bug.
Generic bug implementation for s390. Will increase the value of the console output on BUG() statements since registers r0-r5,r14 will not be clobbered by a printk() call that was previously done before the illegal instruction of BUG() was hit.
Also implements an architecture specific WARN_ON(). Output of that could be increased but requires common code change.
---
arch/s390/Kconfig | 5 ++ arch/s390/defconfig | 2 + arch/s390/kernel/module.c | 4 +- arch/s390/kernel/traps.c | 17 ++++++++-- arch/s390/kernel/vmlinux.lds.S | 10 +++++ include/asm-s390/bug.h | 69 +++++++++++++++++++++++++++++++++-------- 6 files changed, 89 insertions(+), 18 deletions(-)
arch/s390/defconfig +2 -0
@@ -12,6 +12,7 @@ CONFIG_RWSEM_XCHGADD_ALGORITHM=y # CONFIG_ARCH_HAS_ILOG2_U64 is not set CONFIG_GENERIC_HWEIGHT=y CONFIG_GENERIC_TIME=y +CONFIG_GENERIC_BUG=y CONFIG_NO_IOMEM=y CONFIG_S390=y CONFIG_DEFCONFIG_LIST="/lib/modules/$UNAME_RELEASE/.config"
@@ -705,6 +706,7 @@ CONFIG_DEBUG_MUTEXES=y CONFIG_DEBUG_SPINLOCK_SLEEP=y # CONFIG_DEBUG_LOCKING_API_SELFTESTS is not set # CONFIG_DEBUG_KOBJECT is not set +CONFIG_DEBUG_BUGVERBOSE=y # CONFIG_DEBUG_INFO is not set # CONFIG_DEBUG_VM is not set # CONFIG_DEBUG_LIST is not set
arch/s390/Kconfig +5 -0
@@ -41,6 +41,11 @@ config GENERIC_HWEIGHT config GENERIC_TIME def_bool y +config GENERIC_BUG + bool + depends on BUG + default y + config NO_IOMEM def_bool y
arch/s390/kernel/module.c +3 -1
@@ -31,6 +31,7 @@#include <linux/string.h> #include <linux/kernel.h> #include <linux/moduleloader.h> +#include <linux/bug.h> #if 0 #define DEBUGP printk
@@ -398,9 +399,10 @@ int module_finalize(const Elf_Ehdr *hdr, struct module *me) {vfree (me->arch.syminfo); -return 0 ; +return module_bug_finalize (hdr, sechdrs, me); }void module_arch_cleanup (struct module *mod) { +module_bug_cleanup (mod); }
arch/s390/kernel/traps.c +14 -3
@@ -30,7 +30,7 @@#include <linux/kallsyms.h> #include <linux/reboot.h> #include <linux/kprobes.h> - +#include <linux/bug.h> #include <asm/system.h> #include <asm/uaccess.h> #include <asm/io.h>
@@ -297,6 +297,11 @@ report_user_fault(long interruption_code#endif } +int is_valid_bugaddr (unsigned long addr) +{ +return 1 ; +} +static void __kprobes inline do_trap (long interruption_code,int signr,char *str,struct pt_regs *regs,siginfo_t *info)
@@ -323,8 +328,14 @@ static void __kprobes inline do_trap(lon fixup =search_exception_tables (regs->psw.addr &PSW_ADDR_INSN );if (fixup) regs->psw.addr = fixup->fixup |PSW_ADDR_AMODE ; -else -die (str, regs, interruption_code); +else { +enum bug_trap_type btt; + + btt =report_bug (regs->psw.addr &PSW_ADDR_INSN ); +if (btt ==BUG_TRAP_TYPE_WARN ) +return ; +die (str, regs, interruption_code); + } } }
arch/s390/kernel/vmlinux.lds.S +9 -1
@@ -45,6 +45,8 @@ SECTIONS __ex_table : { *(__ex_table) } __stop___ex_table = .; + BUG_TABLE + .data : { /* Data */ *(.data) CONSTRUCTORS
@@ -77,6 +79,12 @@ SECTIONS *(.init.text) _einittext = .; } + /* + * .exit.text is discarded at runtime, not link time, + * to deal with references from __bug_table + */ + .exit.text : { *(.exit.text) } + .init.data : { *(.init.data) } . = ALIGN(256); __setup_start = .;
@@ -116,7 +124,7 @@ SECTIONS /* Sections to be discarded */ /DISCARD/ : { - *(.exit.text) *(.exit.data) *(.exitcall.exit) + *(.exit.data) *(.exitcall.exit) } /* Stabs debugging sections. */
include/asm-s390/bug.h +56 -13
83 lines not on this page · raw message
From: Martin Schwidefsky <[email protected]>
Date: Mon, 5 Mar 2007 23:43:54 +0100
include/asm-s390/bug.h line 56
Quoted 3 lines
include/asm-s390/bug.h +3
-0+if (__builtin_constant_p (__ret_warn_on)) { \ +if (__ret_warn_on) \ +__EMIT_BUG (BUGFLAG_WARNING ); \
I see we'll have this construct on powerpc, parisc and now s390.
But if it's going to trigger essentially at compile time, I think it's much better to BUILD_BUG_ON() in this case instead of counting on the code path to actually run and the user to notice and report the kernel log message.
Quoted 9 lines · David Miller
From: Martin Schwidefsky <[email protected]>
Date: Mon, 5 Mar 2007 23:43:54 +0100include/asm-s390/bug.h line 56
Quoted 3 lines
include/asm-s390/bug.h +3
-0+if (__builtin_constant_p (__ret_warn_on)) { \ +if (__ret_warn_on) \ +__EMIT_BUG (BUGFLAG_WARNING ); \I see we'll have this construct on powerpc, parisc and now s390.
Quoted 4 lines
But if it's going to trigger essentially at compile time, I think it's much better to
BUILD_BUG_ON()in this case instead of counting on the code path to actually run and the user to notice and report the kernel log message.
So something like WARN_ON(1) won't compile, but BUG_ON(1) still would? Seems odd to me.
Also since there is nothing like WARN(), you have to use WARN_ON(1). Btw.: sparc64 has plenty of these ;)
I would prefer to have something like a per arch defined WARN() and a common code WARN_ON() like we have it already for BUG_ON(). Then I would agree: if WARN_ON() is used with a constant the build should fail. Just to tell the author that he did something wrong or simply should use WARN(). And of course the existing BUG_ON() could be converted as well.
From: Heiko Carstens <[email protected]>
Date: Tue, 6 Mar 2007 12:59:17 +0100
Quoted 15 lines · David Miller
Quoted 9 lines
From: Martin Schwidefsky <[email protected]>
Date: Mon, 5 Mar 2007 23:43:54 +0100include/asm-s390/bug.h line 56
Quoted 3 lines
include/asm-s390/bug.h +3
-0+if (__builtin_constant_p (__ret_warn_on)) { \ +if (__ret_warn_on) \ +__EMIT_BUG (BUGFLAG_WARNING ); \I see we'll have this construct on powerpc, parisc and now s390.
Quoted 4 lines
But if it's going to trigger essentially at compile time, I think it's much better to
BUILD_BUG_ON()in this case instead of counting on the code path to actually run and the user to notice and report the kernel log message.
Quoted 4 lines
So something like
WARN_ON(1)won't compile, butBUG_ON(1)still would? Seems odd to me.
Also since there is nothing likeWARN(), you have to useWARN_ON(1). Btw.: sparc64 has plenty of these ;)
I see. It's for case where we can't put the test into the WARN_ON() call.
Thanks for the explanation, I hadn't considered such cases.