* [LTP] [PATCH] getcpu01: Fix strtoul incorrectly returns 0
@ 2023-03-20 9:23 Ping Fang
2023-03-21 6:56 ` Li Wang
0 siblings, 1 reply; 5+ messages in thread
From: Ping Fang @ 2023-03-20 9:23 UTC (permalink / raw)
To: ltp; +Cc: shichen
get_nodeid() find nodeid by iterate /sys/devices/system/node/nodeX/cpuY.
But there are cpulist and cpumap, which will report 0 in strtoul.
On multi-node numa system, the last iteration nodeX directory mismatch
the cpu0 node id. Then fail the test. The following shows on a 4 node
numa system. The last iteration stopped under /node1. so expected node
id 1. Then failed.
getcpu01.c:98: TINFO: cpu:0, cpu_id:0, de->d_name:node2
getcpu01.c:99: TINFO: dn->d_name:cpulist
getcpu01.c:98: TINFO: cpu:0, cpu_id:0, de->d_name:node0
getcpu01.c:99: TINFO: dn->d_name:cpulist
getcpu01.c:98: TINFO: cpu:0, cpu_id:0, de->d_name:node3
getcpu01.c:99: TINFO: dn->d_name:cpulist
getcpu01.c:98: TINFO: cpu:0, cpu_id:0, de->d_name:node1
getcpu01.c:99: TINFO: dn->d_name:cpulist
getcpu01.c:128: TFAIL: getcpu() returned wrong value expected node id:1
returned node id:0, cpu_set:0
Reported-by: Shizhao Chen <shichen@redhat.com>
Signed-off-by: Ping Fang <pifang@redhat.com>
---
testcases/kernel/syscalls/getcpu/getcpu01.c | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/testcases/kernel/syscalls/getcpu/getcpu01.c b/testcases/kernel/syscalls/getcpu/getcpu01.c
index f6fcc4fc1..9842c8999 100644
--- a/testcases/kernel/syscalls/getcpu/getcpu01.c
+++ b/testcases/kernel/syscalls/getcpu/getcpu01.c
@@ -69,6 +69,7 @@ static unsigned int get_nodeid(unsigned int cpu_id)
DIR *directory_parent, *directory_node;
struct dirent *de, *dn;
char directory_path[PATH_MAX];
+ char *invalid_number;
unsigned int cpu;
int node_id = 0;
@@ -91,7 +92,9 @@ static unsigned int get_nodeid(unsigned int cpu_id)
while ((dn = readdir(directory_node)) != NULL) {
if (strncmp(dn->d_name, "cpu", 3))
continue;
- cpu = strtoul(dn->d_name + 3, NULL, 0);
+ cpu = strtoul(dn->d_name + 3, &invalid_number, 0);
+ if (strcmp(invalid_number, "\0"))
+ continue;
if (cpu == cpu_id) {
node_id =
strtoul(de->d_name + 4, NULL, 0);
--
2.31.1
--
Mailing list info: https://lists.linux.it/listinfo/ltp
^ permalink raw reply related [flat|nested] 5+ messages in thread* Re: [LTP] [PATCH] getcpu01: Fix strtoul incorrectly returns 0
2023-03-20 9:23 [LTP] [PATCH] getcpu01: Fix strtoul incorrectly returns 0 Ping Fang
@ 2023-03-21 6:56 ` Li Wang
2023-03-21 11:27 ` Ping Fang
0 siblings, 1 reply; 5+ messages in thread
From: Li Wang @ 2023-03-21 6:56 UTC (permalink / raw)
To: Ping Fang; +Cc: shichen, ltp
Hi Ping,
Good catch!
On Mon, Mar 20, 2023 at 5:23 PM Ping Fang <pifang@redhat.com> wrote:
> get_nodeid() find nodeid by iterate /sys/devices/system/node/nodeX/cpuY.
> But there are cpulist and cpumap, which will report 0 in strtoul.
> On multi-node numa system, the last iteration nodeX directory mismatch
> the cpu0 node id. Then fail the test. The following shows on a 4 node
> numa system. The last iteration stopped under /node1. so expected node
> id 1. Then failed.
>
> getcpu01.c:98: TINFO: cpu:0, cpu_id:0, de->d_name:node2
> getcpu01.c:99: TINFO: dn->d_name:cpulist
> getcpu01.c:98: TINFO: cpu:0, cpu_id:0, de->d_name:node0
> getcpu01.c:99: TINFO: dn->d_name:cpulist
> getcpu01.c:98: TINFO: cpu:0, cpu_id:0, de->d_name:node3
> getcpu01.c:99: TINFO: dn->d_name:cpulist
> getcpu01.c:98: TINFO: cpu:0, cpu_id:0, de->d_name:node1
> getcpu01.c:99: TINFO: dn->d_name:cpulist
> getcpu01.c:128: TFAIL: getcpu() returned wrong value expected node id:1
> returned node id:0, cpu_set:0
>
> Reported-by: Shizhao Chen <shichen@redhat.com>
> Signed-off-by: Ping Fang <pifang@redhat.com>
> ---
> testcases/kernel/syscalls/getcpu/getcpu01.c | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/testcases/kernel/syscalls/getcpu/getcpu01.c
> b/testcases/kernel/syscalls/getcpu/getcpu01.c
> index f6fcc4fc1..9842c8999 100644
> --- a/testcases/kernel/syscalls/getcpu/getcpu01.c
> +++ b/testcases/kernel/syscalls/getcpu/getcpu01.c
> @@ -69,6 +69,7 @@ static unsigned int get_nodeid(unsigned int cpu_id)
> DIR *directory_parent, *directory_node;
> struct dirent *de, *dn;
> char directory_path[PATH_MAX];
> + char *invalid_number;
> unsigned int cpu;
> int node_id = 0;
>
> @@ -91,7 +92,9 @@ static unsigned int get_nodeid(unsigned int cpu_id)
> while ((dn = readdir(directory_node)) != NULL) {
> if (strncmp(dn->d_name, "cpu", 3))
> continue;
> - cpu = strtoul(dn->d_name + 3, NULL, 0);
> + cpu = strtoul(dn->d_name + 3,
> &invalid_number, 0);
> + if (strcmp(invalid_number, "\0"))
> + continue;
>
Why only check if **endptr is '\0' on return? Shouldn't we consider
a more generic situation that was no digits and *endptr stores the
original string (or returns 0)?
> if (cpu == cpu_id) {
> node_id =
> strtoul(de->d_name + 4, NULL,
> 0);
> --
> 2.31.1
>
>
> --
> Mailing list info: https://lists.linux.it/listinfo/ltp
>
>
--
Regards,
Li Wang
--
Mailing list info: https://lists.linux.it/listinfo/ltp
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: [LTP] [PATCH] getcpu01: Fix strtoul incorrectly returns 0
2023-03-21 6:56 ` Li Wang
@ 2023-03-21 11:27 ` Ping Fang
2023-03-21 12:32 ` Li Wang
0 siblings, 1 reply; 5+ messages in thread
From: Ping Fang @ 2023-03-21 11:27 UTC (permalink / raw)
To: Li Wang; +Cc: shichen, ltp
On Tue, Mar 21, 2023 at 2:56 PM Li Wang <liwang@redhat.com> wrote:
> Hi Ping,
>
> Good catch!
>
> On Mon, Mar 20, 2023 at 5:23 PM Ping Fang <pifang@redhat.com> wrote:
>
>> get_nodeid() find nodeid by iterate /sys/devices/system/node/nodeX/cpuY.
>> But there are cpulist and cpumap, which will report 0 in strtoul.
>> On multi-node numa system, the last iteration nodeX directory mismatch
>> the cpu0 node id. Then fail the test. The following shows on a 4 node
>> numa system. The last iteration stopped under /node1. so expected node
>> id 1. Then failed.
>>
>> getcpu01.c:98: TINFO: cpu:0, cpu_id:0, de->d_name:node2
>> getcpu01.c:99: TINFO: dn->d_name:cpulist
>> getcpu01.c:98: TINFO: cpu:0, cpu_id:0, de->d_name:node0
>> getcpu01.c:99: TINFO: dn->d_name:cpulist
>> getcpu01.c:98: TINFO: cpu:0, cpu_id:0, de->d_name:node3
>> getcpu01.c:99: TINFO: dn->d_name:cpulist
>> getcpu01.c:98: TINFO: cpu:0, cpu_id:0, de->d_name:node1
>> getcpu01.c:99: TINFO: dn->d_name:cpulist
>> getcpu01.c:128: TFAIL: getcpu() returned wrong value expected node id:1
>> returned node id:0, cpu_set:0
>>
>> Reported-by: Shizhao Chen <shichen@redhat.com>
>> Signed-off-by: Ping Fang <pifang@redhat.com>
>> ---
>> testcases/kernel/syscalls/getcpu/getcpu01.c | 5 ++++-
>> 1 file changed, 4 insertions(+), 1 deletion(-)
>>
>> diff --git a/testcases/kernel/syscalls/getcpu/getcpu01.c
>> b/testcases/kernel/syscalls/getcpu/getcpu01.c
>> index f6fcc4fc1..9842c8999 100644
>> --- a/testcases/kernel/syscalls/getcpu/getcpu01.c
>> +++ b/testcases/kernel/syscalls/getcpu/getcpu01.c
>> @@ -69,6 +69,7 @@ static unsigned int get_nodeid(unsigned int cpu_id)
>> DIR *directory_parent, *directory_node;
>> struct dirent *de, *dn;
>> char directory_path[PATH_MAX];
>> + char *invalid_number;
>> unsigned int cpu;
>> int node_id = 0;
>>
>> @@ -91,7 +92,9 @@ static unsigned int get_nodeid(unsigned int cpu_id)
>> while ((dn = readdir(directory_node)) != NULL) {
>> if (strncmp(dn->d_name, "cpu", 3))
>> continue;
>> - cpu = strtoul(dn->d_name + 3, NULL, 0);
>> + cpu = strtoul(dn->d_name + 3,
>> &invalid_number, 0);
>> + if (strcmp(invalid_number, "\0"))
>> + continue;
>>
>
> Why only check if **endptr is '\0' on return? Shouldn't we consider
> a more generic situation that was no digits and *endptr stores the
> original string (or returns 0)?
>
The strtoul will try to convert string into unsigned long integer.
If successful the endptr will point to a "\0" string return the number,
otherwise to the original string and return 0.
In the case of /sys/devices/system/node/node1/cpulist, endptr is "list\0"
and return 0. The strcmp false skip to the next iteration.
>
>
>> if (cpu == cpu_id) {
>> node_id =
>> strtoul(de->d_name + 4, NULL,
>> 0);
>> --
>> 2.31.1
>>
>>
>> --
>> Mailing list info: https://lists.linux.it/listinfo/ltp
>>
>>
>
> --
> Regards,
> Li Wang
>
--
*--*
Thanks & Regards,
*Ping Fang*
--
Mailing list info: https://lists.linux.it/listinfo/ltp
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: [LTP] [PATCH] getcpu01: Fix strtoul incorrectly returns 0
2023-03-21 11:27 ` Ping Fang
@ 2023-03-21 12:32 ` Li Wang
2023-03-22 2:04 ` Li Wang
0 siblings, 1 reply; 5+ messages in thread
From: Li Wang @ 2023-03-21 12:32 UTC (permalink / raw)
To: Ping Fang; +Cc: shichen, ltp
On Tue, Mar 21, 2023 at 7:27 PM Ping Fang <pifang@redhat.com> wrote:
>
>
> On Tue, Mar 21, 2023 at 2:56 PM Li Wang <liwang@redhat.com> wrote:
>
>> Hi Ping,
>>
>> Good catch!
>>
>> On Mon, Mar 20, 2023 at 5:23 PM Ping Fang <pifang@redhat.com> wrote:
>>
>>> get_nodeid() find nodeid by iterate /sys/devices/system/node/nodeX/cpuY.
>>> But there are cpulist and cpumap, which will report 0 in strtoul.
>>> On multi-node numa system, the last iteration nodeX directory mismatch
>>> the cpu0 node id. Then fail the test. The following shows on a 4 node
>>> numa system. The last iteration stopped under /node1. so expected node
>>> id 1. Then failed.
>>>
>>> getcpu01.c:98: TINFO: cpu:0, cpu_id:0, de->d_name:node2
>>> getcpu01.c:99: TINFO: dn->d_name:cpulist
>>> getcpu01.c:98: TINFO: cpu:0, cpu_id:0, de->d_name:node0
>>> getcpu01.c:99: TINFO: dn->d_name:cpulist
>>> getcpu01.c:98: TINFO: cpu:0, cpu_id:0, de->d_name:node3
>>> getcpu01.c:99: TINFO: dn->d_name:cpulist
>>> getcpu01.c:98: TINFO: cpu:0, cpu_id:0, de->d_name:node1
>>> getcpu01.c:99: TINFO: dn->d_name:cpulist
>>> getcpu01.c:128: TFAIL: getcpu() returned wrong value expected node id:1
>>> returned node id:0, cpu_set:0
>>>
>>> Reported-by: Shizhao Chen <shichen@redhat.com>
>>> Signed-off-by: Ping Fang <pifang@redhat.com>
>>> ---
>>> testcases/kernel/syscalls/getcpu/getcpu01.c | 5 ++++-
>>> 1 file changed, 4 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/testcases/kernel/syscalls/getcpu/getcpu01.c
>>> b/testcases/kernel/syscalls/getcpu/getcpu01.c
>>> index f6fcc4fc1..9842c8999 100644
>>> --- a/testcases/kernel/syscalls/getcpu/getcpu01.c
>>> +++ b/testcases/kernel/syscalls/getcpu/getcpu01.c
>>> @@ -69,6 +69,7 @@ static unsigned int get_nodeid(unsigned int cpu_id)
>>> DIR *directory_parent, *directory_node;
>>> struct dirent *de, *dn;
>>> char directory_path[PATH_MAX];
>>> + char *invalid_number;
>>> unsigned int cpu;
>>> int node_id = 0;
>>>
>>> @@ -91,7 +92,9 @@ static unsigned int get_nodeid(unsigned int cpu_id)
>>> while ((dn = readdir(directory_node)) != NULL) {
>>> if (strncmp(dn->d_name, "cpu", 3))
>>> continue;
>>> - cpu = strtoul(dn->d_name + 3, NULL, 0);
>>> + cpu = strtoul(dn->d_name + 3,
>>> &invalid_number, 0);
>>> + if (strcmp(invalid_number, "\0"))
>>> + continue;
>>>
>>
>> Why only check if **endptr is '\0' on return? Shouldn't we consider
>> a more generic situation that was no digits and *endptr stores the
>> original string (or returns 0)?
>>
>
>
> The strtoul will try to convert string into unsigned long integer.
> If successful the endptr will point to a "\0" string return the number,
> otherwise to the original string and return 0.
> In the case of /sys/devices/system/node/node1/cpulist, endptr is "list\0"
> and return 0. The strcmp false skip to the next iteration.
>
Ah, you're right. Here we're expecting to get a valid cpuid, the
'dn->d_name + 3' should be strictly equal to "N\0".
Thanks for the explanation.
Reviewed-by: Li Wang <liwang@redhat.com>
--
Regards,
Li Wang
--
Mailing list info: https://lists.linux.it/listinfo/ltp
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2023-03-22 2:04 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2023-03-20 9:23 [LTP] [PATCH] getcpu01: Fix strtoul incorrectly returns 0 Ping Fang
2023-03-21 6:56 ` Li Wang
2023-03-21 11:27 ` Ping Fang
2023-03-21 12:32 ` Li Wang
2023-03-22 2:04 ` Li Wang
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox