[20.08] telemetry: add upper limit on connections
Checks
Commit Message
This patch limits the number of client connections to the new telemetry
socket. The limit is set at 10.
Signed-off-by: Ciara Power <ciara.power@intel.com>
---
lib/librte_telemetry/telemetry.c | 10 ++++++++++
1 file changed, 10 insertions(+)
Comments
On Mon, 18 May 2020 17:12:32 +0100
Ciara Power <ciara.power@intel.com> wrote:
> This patch limits the number of client connections to the new telemetry
> socket. The limit is set at 10.
>
> Signed-off-by: Ciara Power <ciara.power@intel.com>
The upper limit is fine, but using System V semaphores is not a good
way to do this. Please consider using some other atomic primitive.
19/05/2020 02:07, Stephen Hemminger:
> On Mon, 18 May 2020 17:12:32 +0100
> Ciara Power <ciara.power@intel.com> wrote:
>
> > This patch limits the number of client connections to the new telemetry
> > socket. The limit is set at 10.
> >
> > Signed-off-by: Ciara Power <ciara.power@intel.com>
>
> The upper limit is fine, but using System V semaphores is not a good
> way to do this. Please consider using some other atomic primitive.
Why no reply?
>-----Original Message-----
>From: Thomas Monjalon <thomas@monjalon.net>
>Sent: Friday 10 July 2020 21:51
>To: Power, Ciara <ciara.power@intel.com>
>Cc: dev@dpdk.org; Laatz, Kevin <kevin.laatz@intel.com>; Stephen
>Hemminger <stephen@networkplumber.org>; Richardson, Bruce
><bruce.richardson@intel.com>
>Subject: Re: [dpdk-dev] [PATCH 20.08] telemetry: add upper limit on
>connections
>
>19/05/2020 02:07, Stephen Hemminger:
>> On Mon, 18 May 2020 17:12:32 +0100
>> Ciara Power <ciara.power@intel.com> wrote:
>>
>> > This patch limits the number of client connections to the new
>> > telemetry socket. The limit is set at 10.
>> >
>> > Signed-off-by: Ciara Power <ciara.power@intel.com>
>>
>> The upper limit is fine, but using System V semaphores is not a good
>> way to do this. Please consider using some other atomic primitive.
>
>Why no reply?
>
Hi Thomas,
I had contacted Stephen offline for some detail on his suggestion. I thought I would have a patch sent soon after, but it took longer than expected, and I forgot to reply on list.
I have a patch in progress that will hopefully be sent up soon after internal review.
Thanks,
Ciara
@@ -7,6 +7,7 @@
#include <sys/socket.h>
#include <sys/un.h>
#include <dlfcn.h>
+#include <semaphore.h>
/* we won't link against libbsd, so just always use DPDKs-specific strlcpy */
#undef RTE_USE_LIBBSD
@@ -23,6 +24,7 @@
#define MAX_CMD_LEN 56
#define MAX_HELP_LEN 64
#define MAX_OUTPUT_LEN (1024 * 16)
+#define MAX_CONNECTIONS 10
static void *
client_handler(void *socket);
@@ -37,6 +39,7 @@ struct socket {
int sock;
char path[sizeof(((struct sockaddr_un *)0)->sun_path)];
handler fn;
+ sem_t *limit;
};
static struct socket v2_socket; /* socket for v2 telemetry */
static struct socket v1_socket; /* socket for v1 telemetry */
@@ -46,6 +49,7 @@ static struct cmd_callback callbacks[TELEMETRY_MAX_CALLBACKS];
static int num_callbacks; /* How many commands are registered */
/* Used when accessing or modifying list of command callbacks */
static rte_spinlock_t callback_sl = RTE_SPINLOCK_INITIALIZER;
+static sem_t v2_limit;
int
rte_telemetry_register_cmd(const char *cmd, telemetry_cb fn, const char *help)
@@ -263,6 +267,7 @@ client_handler(void *sock_id)
bytes = read(s, buffer, sizeof(buffer) - 1);
}
close(s);
+ sem_post(&v2_limit);
return NULL;
}
@@ -272,6 +277,8 @@ socket_listener(void *socket)
while (1) {
pthread_t th;
struct socket *s = (struct socket *)socket;
+ if (s->limit != NULL)
+ sem_wait(s->limit);
int s_accepted = accept(s->sock, NULL, NULL);
if (s_accepted < 0) {
snprintf(telemetry_log_error,
@@ -372,6 +379,9 @@ telemetry_v2_init(const char *runtime_dir)
{
pthread_t t_new;
+ RTE_BUILD_BUG_ON(_SC_SEM_VALUE_MAX < MAX_CONNECTIONS);
+ (void)sem_init(&v2_limit, 0, MAX_CONNECTIONS);
+ v2_socket.limit = &v2_limit;
rte_telemetry_register_cmd("/", list_commands,
"Returns list of available commands, Takes no parameters");
rte_telemetry_register_cmd("/info", json_info,