[PATCH bpf] tools: bpf: fix NULL return handling in bpf__prepare_load

[PATCH bpf] tools: bpf: fix NULL return handling in bpf__prepare_load

[YueHaibing]

bpf_object__open()/bpf_object__open_buffer can return error pointer or NULL,
check the return values with IS_ERR_OR_NULL() in bpf__prepare_load and
bpf__prepare_load_buffer

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
---
 tools/perf/util/bpf-loader.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/tools/perf/util/bpf-loader.c b/tools/perf/util/bpf-loader.c
index af7ad81..cee6587 100644
--- a/tools/perf/util/bpf-loader.c
+++ b/tools/perf/util/bpf-loader.c
@@ -66,7 +66,7 @@ bpf__prepare_load_buffer(void *obj_buf, size_t obj_buf_sz, const char *name)
 	}
 
 	obj = bpf_object__open_buffer(obj_buf, obj_buf_sz, name);
-	if (IS_ERR(obj)) {
+	if (IS_ERR_OR_NULL(obj)) {
 		pr_debug("bpf: failed to load buffer\n");
 		return ERR_PTR(-EINVAL);
 	}
@@ -102,14 +102,14 @@ struct bpf_object *bpf__prepare_load(const char *filename, bool source)
 			pr_debug("bpf: successfull builtin compilation\n");
 		obj = bpf_object__open_buffer(obj_buf, obj_buf_sz, filename);
 
-		if (!IS_ERR(obj) && llvm_param.dump_obj)
+		if (!IS_ERR_OR_NULL(obj) && llvm_param.dump_obj)
 			llvm__dump_obj(filename, obj_buf, obj_buf_sz);
 
 		free(obj_buf);
 	} else
 		obj = bpf_object__open(filename);
 
-	if (IS_ERR(obj)) {
+	if (IS_ERR_OR_NULL(obj)) {
 		pr_debug("bpf: failed to load %s\n", filename);
 		return obj;
 	}
-- 
2.7.0

Re: [PATCH bpf] tools: bpf: fix NULL return handling in bpf__prepare_load

[Daniel Borkmann]

[ +Arnaldo ]

On 05/11/2018 01:21 PM, YueHaibing wrote:
> bpf_object__open()/bpf_object__open_buffer can return error pointer or NULL,
> check the return values with IS_ERR_OR_NULL() in bpf__prepare_load and
> bpf__prepare_load_buffer
> 
> Signed-off-by: YueHaibing <yuehaibing@huawei.com>
> ---
>  tools/perf/util/bpf-loader.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)

This should probably be routed via Arnaldo due to the fix in perf itself. If
there's no particular preference on which tree, we could potentially route it
as well via bpf with Acked-by from Arnaldo, but that is up to him. Arnaldo,
any preference?

> diff --git a/tools/perf/util/bpf-loader.c b/tools/perf/util/bpf-loader.c
> index af7ad81..cee6587 100644
> --- a/tools/perf/util/bpf-loader.c
> +++ b/tools/perf/util/bpf-loader.c
> @@ -66,7 +66,7 @@ bpf__prepare_load_buffer(void *obj_buf, size_t obj_buf_sz, const char *name)
>  	}
>  
>  	obj = bpf_object__open_buffer(obj_buf, obj_buf_sz, name);
> -	if (IS_ERR(obj)) {
> +	if (IS_ERR_OR_NULL(obj)) {
>  		pr_debug("bpf: failed to load buffer\n");
>  		return ERR_PTR(-EINVAL);
>  	}
> @@ -102,14 +102,14 @@ struct bpf_object *bpf__prepare_load(const char *filename, bool source)
>  			pr_debug("bpf: successfull builtin compilation\n");
>  		obj = bpf_object__open_buffer(obj_buf, obj_buf_sz, filename);
>  
> -		if (!IS_ERR(obj) && llvm_param.dump_obj)
> +		if (!IS_ERR_OR_NULL(obj) && llvm_param.dump_obj)
>  			llvm__dump_obj(filename, obj_buf, obj_buf_sz);
>  
>  		free(obj_buf);
>  	} else
>  		obj = bpf_object__open(filename);
>  
> -	if (IS_ERR(obj)) {
> +	if (IS_ERR_OR_NULL(obj)) {
>  		pr_debug("bpf: failed to load %s\n", filename);
>  		return obj;
>  	}
> 

Re: [PATCH bpf] tools: bpf: fix NULL return handling in bpf__prepare_load

[Arnaldo Carvalho de Melo]

Em Sun, May 13, 2018 at 01:20:22AM +0200, Daniel Borkmann escreveu:
> [ +Arnaldo ]
> 
> On 05/11/2018 01:21 PM, YueHaibing wrote:
> > bpf_object__open()/bpf_object__open_buffer can return error pointer or NULL,
> > check the return values with IS_ERR_OR_NULL() in bpf__prepare_load and
> > bpf__prepare_load_buffer
> > 
> > Signed-off-by: YueHaibing <yuehaibing@huawei.com>
> > ---
> >  tools/perf/util/bpf-loader.c | 6 +++---
> >  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> This should probably be routed via Arnaldo due to the fix in perf itself. If
> there's no particular preference on which tree, we could potentially route it
> as well via bpf with Acked-by from Arnaldo, but that is up to him. Arnaldo,
> any preference?

I'm preparing a pull req right now, and working a bit on perf's BPF
support, so why not, I'll merge it, thanks,

- Arnaldo
 
> > diff --git a/tools/perf/util/bpf-loader.c b/tools/perf/util/bpf-loader.c
> > index af7ad81..cee6587 100644
> > --- a/tools/perf/util/bpf-loader.c
> > +++ b/tools/perf/util/bpf-loader.c
> > @@ -66,7 +66,7 @@ bpf__prepare_load_buffer(void *obj_buf, size_t obj_buf_sz, const char *name)
> >  	}
> >  
> >  	obj = bpf_object__open_buffer(obj_buf, obj_buf_sz, name);
> > -	if (IS_ERR(obj)) {
> > +	if (IS_ERR_OR_NULL(obj)) {
> >  		pr_debug("bpf: failed to load buffer\n");
> >  		return ERR_PTR(-EINVAL);
> >  	}
> > @@ -102,14 +102,14 @@ struct bpf_object *bpf__prepare_load(const char *filename, bool source)
> >  			pr_debug("bpf: successfull builtin compilation\n");
> >  		obj = bpf_object__open_buffer(obj_buf, obj_buf_sz, filename);
> >  
> > -		if (!IS_ERR(obj) && llvm_param.dump_obj)
> > +		if (!IS_ERR_OR_NULL(obj) && llvm_param.dump_obj)
> >  			llvm__dump_obj(filename, obj_buf, obj_buf_sz);
> >  
> >  		free(obj_buf);
> >  	} else
> >  		obj = bpf_object__open(filename);
> >  
> > -	if (IS_ERR(obj)) {
> > +	if (IS_ERR_OR_NULL(obj)) {
> >  		pr_debug("bpf: failed to load %s\n", filename);
> >  		return obj;
> >  	}
> > 

Re: [PATCH bpf] tools: bpf: fix NULL return handling in bpf__prepare_load

[Daniel Borkmann]

On 05/15/2018 04:11 PM, Arnaldo Carvalho de Melo wrote:
> Em Sun, May 13, 2018 at 01:20:22AM +0200, Daniel Borkmann escreveu:
>> [ +Arnaldo ]
>>
>> On 05/11/2018 01:21 PM, YueHaibing wrote:
>>> bpf_object__open()/bpf_object__open_buffer can return error pointer or NULL,
>>> check the return values with IS_ERR_OR_NULL() in bpf__prepare_load and
>>> bpf__prepare_load_buffer
>>>
>>> Signed-off-by: YueHaibing <yuehaibing@huawei.com>
>>> ---
>>>  tools/perf/util/bpf-loader.c | 6 +++---
>>>  1 file changed, 3 insertions(+), 3 deletions(-)
>>
>> This should probably be routed via Arnaldo due to the fix in perf itself. If
>> there's no particular preference on which tree, we could potentially route it
>> as well via bpf with Acked-by from Arnaldo, but that is up to him. Arnaldo,
>> any preference?
> 
> I'm preparing a pull req right now, and working a bit on perf's BPF
> support, so why not, I'll merge it, thanks,

Sounds good, thanks Arnaldo!

> - Arnaldo