On 07/03/2020 09:39, Phong Tran wrote: > follow the suggestion from > https://github.com/KSPP/linux/issues/35 > > Signed-off-by: Phong Tran <tranmanphong@xxxxxxxxx> > --- > arch/arm64/Kconfig.debug | 3 ++- > arch/arm64/include/asm/ptdump.h | 2 ++ > arch/arm64/mm/dump.c | 1 + > arch/arm64/mm/ptdump_debugfs.c | 18 ++++++++++++++++++ > 4 files changed, 23 insertions(+), 1 deletion(-) > > diff --git a/arch/arm64/Kconfig.debug b/arch/arm64/Kconfig.debug > index 1c906d932d6b..be552fa351e2 100644 > --- a/arch/arm64/Kconfig.debug > +++ b/arch/arm64/Kconfig.debug > @@ -48,7 +48,8 @@ config DEBUG_WX > of other unfixed kernel bugs easier. > > There is no runtime or memory usage effect of this option > - once the kernel has booted up - it's a one time check. > + once the kernel has booted up - it's a one time check and > + can be checked by echo "1" to "check_wx_pages" debugfs in runtime. I would suggest it may be better spelling this out a bit more, because at the moment it's a little confusing when the config option is "Warn on W+X mappings at boot", but your change makes it sound like it only happens when you do the echo. Perhaps something like: There is no runtime or memory usage effect of this option once the kernel has booted up - it's a one time check at boot, and can also be triggered at runtime by echo "1" to "check_wx_pages". > > If in doubt, say "Y". > > diff --git a/arch/arm64/include/asm/ptdump.h b/arch/arm64/include/asm/ptdump.h > index 38187f74e089..b80d6b4fc508 100644 > --- a/arch/arm64/include/asm/ptdump.h > +++ b/arch/arm64/include/asm/ptdump.h > @@ -24,9 +24,11 @@ struct ptdump_info { > void ptdump_walk(struct seq_file *s, struct ptdump_info *info); > #ifdef CONFIG_PTDUMP_DEBUGFS > void ptdump_debugfs_register(struct ptdump_info *info, const char *name); > +int ptdump_check_wx_init(void); > #else > static inline void ptdump_debugfs_register(struct ptdump_info *info, > const char *name) { } > +static inline int ptdump_check_wx_init(void) { return 0; } > #endif > void ptdump_check_wx(void); > #endif /* CONFIG_PTDUMP_CORE */ This is a confusing! Why have you made it dependent on CONFIG_PTDUMP_DEBUGFS? Well actually I can see why - it's because you've put the new functions in ptdump_debugfs.c which is (currently) only built when CONFIG_PTDUMP_DBEUGFS is enabled. So you need to either: a) Ensure the new code is built when CONFIG_PTDUMP_DEBUGFS isn't enabled. b) Update the Kconfig help text to say that the debugfs file for triggering a runtime W+X check is only available if CONFIG_PTDUMP_DEBUGFS is also enabled. Other than the confusion over config symbols this looks good. Thanks, Steve > diff --git a/arch/arm64/mm/dump.c b/arch/arm64/mm/dump.c > index 860c00ec8bd3..60c99a047763 100644 > --- a/arch/arm64/mm/dump.c > +++ b/arch/arm64/mm/dump.c > @@ -378,6 +378,7 @@ static int ptdump_init(void) > #endif > ptdump_initialize(); > ptdump_debugfs_register(&kernel_ptdump_info, "kernel_page_tables"); > + ptdump_check_wx_init(); > return 0; > } > device_initcall(ptdump_init); > diff --git a/arch/arm64/mm/ptdump_debugfs.c b/arch/arm64/mm/ptdump_debugfs.c > index 1f2eae3e988b..73cddc12c3c2 100644 > --- a/arch/arm64/mm/ptdump_debugfs.c > +++ b/arch/arm64/mm/ptdump_debugfs.c > @@ -16,3 +16,21 @@ void ptdump_debugfs_register(struct ptdump_info *info, const char *name) > { > debugfs_create_file(name, 0400, NULL, info, &ptdump_fops); > } > + > +static int check_wx_debugfs_set(void *data, u64 val) > +{ > + if (val != 1ULL) > + return -EINVAL; > + > + ptdump_check_wx(); > + > + return 0; > +} > + > +DEFINE_SIMPLE_ATTRIBUTE(check_wx_fops, NULL, check_wx_debugfs_set, "%llu\n"); > + > +int ptdump_check_wx_init(void) > +{ > + return debugfs_create_file("check_wx_pages", 0200, NULL, > + NULL, &check_wx_fops) ? 0 : -ENOMEM; > +} >
