[bug report] x86/mce: Prevent severity computation from being instrumented

[bug report] x86/mce: Prevent severity computation from being instrumented

[Dan Carpenter]

Hello Borislav Petkov,

This is a semi-automatic email about new static checker warnings.

The patch 0a5b288e85bb: "x86/mce: Prevent severity computation from
being instrumented" from Oct 13, 2021, leads to the following Smatch
complaint:

    arch/x86/kernel/cpu/mce/severity.c:286 error_context()
    warn: variable dereferenced before check 'regs' (see line 280)

arch/x86/kernel/cpu/mce/severity.c
   279		fixup_type = ex_get_fixup_type(m->ip);
   280		copy_user  = is_copy_from_user(regs);
                                               ^^^^
New unchecked dereference

   281		instrumentation_end();
   282	
   283		switch (fixup_type) {
   284		case EX_TYPE_UACCESS:
   285		case EX_TYPE_COPY:
   286			if (!regs || !copy_user)
                             ^^^^
Old code checked for NULL

   287				return IN_KERNEL;
   288			m->kflags |= MCE_IN_KERNEL_COPYIN;

regards,
dan carpenter

Re: [bug report] x86/mce: Prevent severity computation from being instrumented

[Borislav Petkov]

+ Tony.

On Fri, Dec 17, 2021 at 01:20:29PM +0300, Dan Carpenter wrote:
> Hello Borislav Petkov,
> 
> This is a semi-automatic email about new static checker warnings.
> 
> The patch 0a5b288e85bb: "x86/mce: Prevent severity computation from
> being instrumented" from Oct 13, 2021, leads to the following Smatch
> complaint:
> 
>     arch/x86/kernel/cpu/mce/severity.c:286 error_context()
>     warn: variable dereferenced before check 'regs' (see line 280)
> 
> arch/x86/kernel/cpu/mce/severity.c
>    279		fixup_type = ex_get_fixup_type(m->ip);
>    280		copy_user  = is_copy_from_user(regs);
>                                                ^^^^
> New unchecked dereference
> 
>    281		instrumentation_end();
>    282	
>    283		switch (fixup_type) {
>    284		case EX_TYPE_UACCESS:
>    285		case EX_TYPE_COPY:
>    286			if (!regs || !copy_user)
>                              ^^^^
> Old code checked for NULL
> 
>    287				return IN_KERNEL;
>    288			m->kflags |= MCE_IN_KERNEL_COPYIN;

Good catch, thanks!

I guess we can do something like this:

---
diff --git a/arch/x86/kernel/cpu/mce/severity.c b/arch/x86/kernel/cpu/mce/severity.c
index a32646769705..7aa2bda93cbb 100644
--- a/arch/x86/kernel/cpu/mce/severity.c
+++ b/arch/x86/kernel/cpu/mce/severity.c
@@ -222,6 +222,9 @@ static bool is_copy_from_user(struct pt_regs *regs)
 	struct insn insn;
 	int ret;
 
+	if (!regs)
+		return false;
+
 	if (copy_from_kernel_nofault(insn_buf, (void *)regs->ip, MAX_INSN_SIZE))
 		return false;
 
@@ -283,7 +286,7 @@ static noinstr int error_context(struct mce *m, struct pt_regs *regs)
 	switch (fixup_type) {
 	case EX_TYPE_UACCESS:
 	case EX_TYPE_COPY:
-		if (!regs || !copy_user)
+		if (!copy_user)
 			return IN_KERNEL;
 		m->kflags |= MCE_IN_KERNEL_COPYIN;
 		fallthrough;

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette

Re: [bug report] x86/mce: Prevent severity computation from being instrumented

[Luck, Tony]

On Fri, Dec 17, 2021 at 12:29:45PM +0100, Borislav Petkov wrote:
> + Tony.
> 
> On Fri, Dec 17, 2021 at 01:20:29PM +0300, Dan Carpenter wrote:
> > Hello Borislav Petkov,
> > 
> > This is a semi-automatic email about new static checker warnings.
> > 
> > The patch 0a5b288e85bb: "x86/mce: Prevent severity computation from
> > being instrumented" from Oct 13, 2021, leads to the following Smatch
> > complaint:
> > 
> >     arch/x86/kernel/cpu/mce/severity.c:286 error_context()
> >     warn: variable dereferenced before check 'regs' (see line 280)
> > 
> > arch/x86/kernel/cpu/mce/severity.c
> >    279		fixup_type = ex_get_fixup_type(m->ip);
> >    280		copy_user  = is_copy_from_user(regs);
> >                                                ^^^^
> > New unchecked dereference
> > 
> >    281		instrumentation_end();
> >    282	
> >    283		switch (fixup_type) {
> >    284		case EX_TYPE_UACCESS:
> >    285		case EX_TYPE_COPY:
> >    286			if (!regs || !copy_user)
> >                              ^^^^
> > Old code checked for NULL
> > 
> >    287				return IN_KERNEL;
> >    288			m->kflags |= MCE_IN_KERNEL_COPYIN;
> 
> Good catch, thanks!
> 
> I guess we can do something like this:
> 
> ---
> diff --git a/arch/x86/kernel/cpu/mce/severity.c b/arch/x86/kernel/cpu/mce/severity.c
> index a32646769705..7aa2bda93cbb 100644
> --- a/arch/x86/kernel/cpu/mce/severity.c
> +++ b/arch/x86/kernel/cpu/mce/severity.c
> @@ -222,6 +222,9 @@ static bool is_copy_from_user(struct pt_regs *regs)
>  	struct insn insn;
>  	int ret;
>  
> +	if (!regs)
> +		return false;
> +
>  	if (copy_from_kernel_nofault(insn_buf, (void *)regs->ip, MAX_INSN_SIZE))
>  		return false;
>  
> @@ -283,7 +286,7 @@ static noinstr int error_context(struct mce *m, struct pt_regs *regs)
>  	switch (fixup_type) {
>  	case EX_TYPE_UACCESS:
>  	case EX_TYPE_COPY:
> -		if (!regs || !copy_user)
> +		if (!copy_user)
>  			return IN_KERNEL;
>  		m->kflags |= MCE_IN_KERNEL_COPYIN;
>  		fallthrough;
> 

Looks good to me.

Reviewed-by: Tony Luck <tony.luck@intel.com>

-Tony

[PATCH] x86/mce: Check regs before accessing it

[Borislav Petkov]

From: Borislav Petkov <bp@suse.de>

Commit in Fixes accesses pt_regs before checking whether it is NULL or
not. Change the flow so that the NULL pointer check happens first.

Fixes: 0a5b288e85bb ("x86/mce: Prevent severity computation from being instrumented")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Tony Luck <tony.luck@intel.com>
Link: https://lore.kernel.org/r/20211217102029.GA29708@kili
---
 arch/x86/kernel/cpu/mce/severity.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kernel/cpu/mce/severity.c b/arch/x86/kernel/cpu/mce/severity.c
index a32646769705..7aa2bda93cbb 100644
--- a/arch/x86/kernel/cpu/mce/severity.c
+++ b/arch/x86/kernel/cpu/mce/severity.c
@@ -222,6 +222,9 @@ static bool is_copy_from_user(struct pt_regs *regs)
 	struct insn insn;
 	int ret;
 
+	if (!regs)
+		return false;
+
 	if (copy_from_kernel_nofault(insn_buf, (void *)regs->ip, MAX_INSN_SIZE))
 		return false;
 
@@ -283,7 +286,7 @@ static noinstr int error_context(struct mce *m, struct pt_regs *regs)
 	switch (fixup_type) {
 	case EX_TYPE_UACCESS:
 	case EX_TYPE_COPY:
-		if (!regs || !copy_user)
+		if (!copy_user)
 			return IN_KERNEL;
 		m->kflags |= MCE_IN_KERNEL_COPYIN;
 		fallthrough;
-- 
2.29.2


-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette