Re: [lttng-dev] Test if tracepoint is enabled
On 05/12/12 20:49, Mathieu Desnoyers wrote:
* David Bryant (david.bry...@quantum.com) wrote:
So, building on your example:
#define tracepoint_cond(provider, name, pre, post, ...) \
do { \
if (caa_unlikely(__tracepoint_##provider##___##name.state)) { \
pre \
__tracepoint_cb_##provider##___##name(__VA_ARGS__); \
post \
} \
} \
while (0)
And a contrived example:
#include stdio.h
#include stdlib.h
int main() {
char * a;
tracepoint_cond(provider, name,
{ printf(pre\n); a = malloc(16); },
{ printf(post\n); free(a);}
);
return 0;
}
This is ok. But more work with the macro handling would be required to
get the pre/post code fragments through as parameters. The example only
compiles because its pre/post arguments don't contain any commas.
What do you think?
I'm not sure what parameters would cause issues with commas. If they are
between parenthesis or brackets, it should be fine. Do you have an
example perhaps ?
Ok, I've just learnt what 'statement-expressions' are
(http://gcc.gnu.org/onlinedocs/gcc/Statement-Exprs.html). Would this be
the first use of these in lttng-ust? If so, I'm wary of making lttng-ust
dependent on this gcc extension. Would it be dependent on this? It is up
to the user to decide what to pass to the macro, but the macro may
impose constraints by way of how the arguments are employed within the
macro...
As it happens, we compile with -pedantic which won't allow
statement-expressions. (As an aside, -pedantic doesn't like tracepoint
that don't take any arguments because it expects the ... to have at
least one argument).
My example above (that doesn't use statement-expressions - I removed the
round brackets) compiles correctly. But it you introduce a comma then
the compiler seems to interpret this as an argument separator. So this
works:
#include stdio.h
#include stdlib.h
#define tracepoint_cond(provider, name, cond, pre, post, ...) \
do { \
if (caa_unlikely(__tracepoint_##provider##___##name.state)) { \
if (cond) { \
pre \
__tracepoint_cb_##provider##___##name(__VA_ARGS__); \
post \
} \
} \
} \
while (0)
int main() {
int i = 1, j = 1;
char * a;
tracepoint_cond(provider, name,
i == j, /* condition */
{ /* pre */
a = malloc(16);
memset(a, 0, 16);
},
{ /* post */
free(a);
int dummy1;
},
a
);
return 0;
}
But changing tracepoint_cond to this breaks it:
tracepoint_cond(provider, name,
i == j, /* condition */
{ /* pre */
a = malloc(16);
memset(a, 0, 16);
},
{ /* post */
free(a);
int dummy1, dummy2; /* XXX */
},
a
);
Examining the pre-processed code reveals that the comma causes dummy2 to
become a subsequent argument to the macro and land in the ... region,
and hell breaks loose.
Do you think we could come up with a scheme that doesn't impose the use
of statement-expressions, but supports them if they are provided? In my
example, I simply provide 'i == j' as the condition, but a
statement-expression would also work. Are we conflating separate ideas -
the idea of user conditions and the idea of setup/cleanup?
While we are there, it would be good to make the pre handler test a
condition. For your use-case, you would just have to pass a pre :
{ printf(pre\n); a = malloc(16); 1; }
So we could do:
if (pre)
__tracepoint_cb_##provider##___##name(__VA_ARGS__);
(post)
and therefore support not only to setup something, but also to add an
extra check on whether tracing should be skipped or not. If someone
passes 1 as end of the compound statement, it will always evaluate to
1, and therefore the compiler will remove any extra branch in -O2.
One question stands though: how would we integrate this with sdt.h
STAP_PROVEV() ?
I'm yet to look at this.
Thanks,
Dave
--
The information contained in this transmission may be confidential. Any
disclosure, copying, or further distribution of confidential information is not
permitted unless such privilege is explicitly granted in writing by Quantum.
Quantum reserves the right to have electronic communications, including email
and attachments, sent across its networks filtered through anti virus and spam
software programs and retain such messages in order to
Re: [lttng-dev] Test if tracepoint is enabled
On 05/12/12 00:32, Mathieu Desnoyers wrote:
Hrm. So what you would like is something like:
int tracepoint_get_state(provider, name);
So you could use it for your setup and cleanup. However, let's see the
possible use of this construct:
1 - this construct would be OK:
if (tracepoint_get_state(myprovider, myname)) {
setup();
tracepoint(myprovider, myname, args...);
cleanup();
}
but we see that the state test would happen twice, which is somewhat
inefficient.
But the actual problem comes from this construct, which is _not_ OK, and
racy:
2 -
if (tracepoint_get_state(myprovider, myname))
setup();
tracepoint(myprovider, myname, args...);
if (tracepoint_get_state(myprovider, myname))
cleanup();
The problem is that users would expect that some synchronization
magically happen when enabling/disabling tracepoints, but it's not the
case. So we could very well end up executing just the cleanup, or just
the tracepoint, or 2 of the 3, if we hit the tracepoint while it is
being enabled/disabled.
Providing this level of deep access into the tracepoint structures seems
error-prone (as shown by construct #2), and inefficient (as shown by the
2 tests performed by construct #1).
This is why I am tempted to introduce a tracepoint_cond() or something
similar, to ensure that no racy construct can be created, and ensure
that we only test the tracepoint state once per site.
Thoughts ?
This construct avoids the inefficiency of checking tracepoint state twice:
if (tracepoint_get_state(myprovider, myname) {
setup();
tracepoint_force(myprovider, myname, args...);
cleanup();
}
ie, tracepoint_force() doesn't re-check the state. However, it relies on
the user adhering to the pattern. tracepoint_force() could also be
misused, violating the state of enabled/disabled tracepoints.
I take your point about the user being easily misled. It's not obvious
enough that tracepoints are enabled/disabled asynchronously and we
should only provide safe constructs that are compatible with that model.
So, building on your example:
#define tracepoint_cond(provider, name, pre, post, ...) \
do { \
if (caa_unlikely(__tracepoint_##provider##___##name.state)) { \
pre \
__tracepoint_cb_##provider##___##name(__VA_ARGS__); \
post \
} \
} \
while (0)
And a contrived example:
#include stdio.h
#include stdlib.h
int main() {
char * a;
tracepoint_cond(provider, name,
{ printf(pre\n); a = malloc(16); },
{ printf(post\n); free(a);}
);
return 0;
}
This is ok. But more work with the macro handling would be required to
get the pre/post code fragments through as parameters. The example only
compiles because its pre/post arguments don't contain any commas.
What do you think?
Thanks,
Dave
--
The information contained in this transmission may be confidential. Any
disclosure, copying, or further distribution of confidential information is not
permitted unless such privilege is explicitly granted in writing by Quantum.
Quantum reserves the right to have electronic communications, including email
and attachments, sent across its networks filtered through anti virus and spam
software programs and retain such messages in order to comply with applicable
data security and retention requirements. Quantum is not responsible for the
proper and complete transmission of the substance of this communication or for
any delay in its receipt.
___
lttng-dev mailing list
lttng-dev@lists.lttng.org
http://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
Re: [lttng-dev] Test if tracepoint is enabled
* David Bryant (david.bry...@quantum.com) wrote:
Hi all,
Is it feasible to have an officially supported way of determining if a
tracepoint is currently enabled?
In my code I would like to test if the tracepoint is enabled, and if it
isn't, avoid doing expensive setup work.
I've been doing the following:
if
(__builtin_expect(!!(__tracepoint_sample_component___message.state), 0))
{
/* setup work */
tracepoint(sample_component, message);
/* cleanup work */
}
But it's weak for two reasons:
* The 'state' attribute goes true when the tracepoint is enabled, but
stays true after the tracepoint is disabled. It only goes false when
the session is destroyed.
* It uses a private / unofficial API
What I'd like is an official API that correctly evaluates whether a
tracepoint is enabled or not?
Is this possible?
Something like:
tracepoint_cond(provider, name, cond, ...)
might do it.
Where cond would be an expression (could be a statement-expression),
e.g.:
tracepoint_cond(myprovider, myevent,
({
a = something;
z = other;
a == z;
}),
a, z);
So we could, in addition to perform some setup, also evaluate a
condition before calling the tracepoint. The macro might look like
(without the '\' at EOL):
#define tracepoint_cond(provider, name, cond, ...)
do {
if (caa_unlikely(__tracepoint_##provider##___##name.state)) {
if (cond)
__tracepoint_cb_##provider##___##name(__VA_ARGS__);
}
} while (0)
So in your case, a statement-expression that evaluates to true would
be fine.
Note that adding support for STAP_PROBEV in there might be a bit less
straightforward, because STAP_PROBEV don't depend on the tracepoint
state.
Thoughts ?
Thanks,
Mathieu
Thanks,
Dave
--
The information contained in this transmission may be confidential. Any
disclosure, copying, or further distribution of confidential information is
not permitted unless such privilege is explicitly granted in writing by
Quantum. Quantum reserves the right to have electronic communications,
including email and attachments, sent across its networks filtered through
anti virus and spam software programs and retain such messages in order to
comply with applicable data security and retention requirements. Quantum is
not responsible for the proper and complete transmission of the substance of
this communication or for any delay in its receipt.
___
lttng-dev mailing list
lttng-dev@lists.lttng.org
http://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
--
Mathieu Desnoyers
Operating System Efficiency RD Consultant
EfficiOS Inc.
http://www.efficios.com
___
lttng-dev mailing list
lttng-dev@lists.lttng.org
http://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
Re: [lttng-dev] Test if tracepoint is enabled
On 04/12/12 01:04, Mathieu Desnoyers wrote:
* David Bryant (david.bry...@quantum.com) wrote:
Hi all,
Is it feasible to have an officially supported way of determining if a
tracepoint is currently enabled?
In my code I would like to test if the tracepoint is enabled, and if it
isn't, avoid doing expensive setup work.
I've been doing the following:
if
(__builtin_expect(!!(__tracepoint_sample_component___message.state), 0))
{
/* setup work */
tracepoint(sample_component, message);
/* cleanup work */
}
But it's weak for two reasons:
* The 'state' attribute goes true when the tracepoint is enabled, but
stays true after the tracepoint is disabled. It only goes false when
the session is destroyed.
* It uses a private / unofficial API
What I'd like is an official API that correctly evaluates whether a
tracepoint is enabled or not?
Is this possible?
Something like:
tracepoint_cond(provider, name, cond, ...)
might do it.
Where cond would be an expression (could be a statement-expression),
e.g.:
tracepoint_cond(myprovider, myevent,
({
a = something;
z = other;
a == z;
}),
a, z);
So we could, in addition to perform some setup, also evaluate a
condition before calling the tracepoint. The macro might look like
(without the '\' at EOL):
#define tracepoint_cond(provider, name, cond, ...)
do {
if (caa_unlikely(__tracepoint_##provider##___##name.state)) {
if (cond)
__tracepoint_cb_##provider##___##name(__VA_ARGS__);
}
} while (0)
Hi Mathieu,
Your proposal seems to introduce a way of providing an additional
condition for whether to trigger the tracepoint, ie, not only would the
tracepoint need to be enabled but also 'cond' must be satisfied. This is
cool, but what I want right now is simply a way of determining whether a
tracepoint is enabled. If it isn't enabled then I won't do the
setup/cleanup work to prepare the arguments.
I've been consulting __tracepoint_##provider##___##name.state but am
unsatisfied because it's unofficial API and it isn't quite the right test.
Thanks,
Dave
So in your case, a statement-expression that evaluates to true would
be fine.
Note that adding support for STAP_PROBEV in there might be a bit less
straightforward, because STAP_PROBEV don't depend on the tracepoint
state.
Thoughts ?
Thanks,
Mathieu
Thanks,
Dave
--
The information contained in this transmission may be confidential. Any
disclosure, copying, or further distribution of confidential information is not
permitted unless such privilege is explicitly granted in writing by Quantum.
Quantum reserves the right to have electronic communications, including email
and attachments, sent across its networks filtered through anti virus and spam
software programs and retain such messages in order to comply with applicable
data security and retention requirements. Quantum is not responsible for the
proper and complete transmission of the substance of this communication or for
any delay in its receipt.
___
lttng-dev mailing list
lttng-dev@lists.lttng.org
http://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
___
lttng-dev mailing list
lttng-dev@lists.lttng.org
http://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
