[ECOS] posix time functions

[ECOS] posix time functions

[David Brennan]

We discovered a (perhaps) un-documented feature in the eCos posix 
clock_settime and clock_gettime functions. Setting the time, changes the 
system clock (number of ticks). The system clock should only be 
changeable by the interrupt routine which is incrementing it. All other 
code should really only have read access. Is there a chance this could 
be changed? That is, if we fix it in the kernel, would that type of 
patch be accepted. I would think that setting the clock and possibly 
losing alarms would be something that other people have noticed before. 
For now we have implemented our own time off-set and never call settime 
and call our own gettime which adds our offset to the time that the OS 
is returning. But this seems like the hard way of doing it.

David Brennan

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] posix time functions

[Nick Garnett]

David Brennan <eCos@brennanhome.com> writes:

> We discovered a (perhaps) un-documented feature in the eCos posix
> clock_settime and clock_gettime functions. Setting the time, changes
> the system clock (number of ticks). The system clock should only be
> changeable by the interrupt routine which is incrementing it. All
> other code should really only have read access.

Hmm, we have also had complaints that clock_settime() doesn't change
enough and want it to alter the wallclock time as well. I guess you
cannot please everyone.

> Is there a chance this
> could be changed? That is, if we fix it in the kernel, would that type
> of patch be accepted.

A patch that maintains an offet for CLOCK_REALTIME from the main
timer and used that when calculating timer timeouts would be
fine. Depending on how much gets changed, we might need an assignment
for it, though.

> I would think that setting the clock and
> possibly losing alarms would be something that other people have
> noticed before.

To be honest, I don't think that many people call clock_settime().
It's mainly there for completeness.

-- 
Nick Garnett                    eCos Kernel Architect
http://www.ecoscentric.com/     The eCos and RedBoot experts


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] posix time functions

[Sam Pham]

Our application has the need of calling
clock_settime() to set the time.  What are the
consequences of calling this function?

Thanks.

Sam


--- Nick Garnett <nickg@ecoscentric.com> wrote:

> David Brennan <eCos@brennanhome.com> writes:
> 
> > We discovered a (perhaps) un-documented feature in
> the eCos posix
> > clock_settime and clock_gettime functions. Setting
> the time, changes
> > the system clock (number of ticks). The system
> clock should only be
> > changeable by the interrupt routine which is
> incrementing it. All
> > other code should really only have read access.
> 
> Hmm, we have also had complaints that
> clock_settime() doesn't change
> enough and want it to alter the wallclock time as
> well. I guess you
> cannot please everyone.
> 
> > Is there a chance this
> > could be changed? That is, if we fix it in the
> kernel, would that type
> > of patch be accepted.
> 
> A patch that maintains an offet for CLOCK_REALTIME
> from the main
> timer and used that when calculating timer timeouts
> would be
> fine. Depending on how much gets changed, we might
> need an assignment
> for it, though.
> 
> > I would think that setting the clock and
> > possibly losing alarms would be something that
> other people have
> > noticed before.
> 
> To be honest, I don't think that many people call
> clock_settime().
> It's mainly there for completeness.
> 
> -- 
> Nick Garnett                    eCos Kernel
> Architect
> http://www.ecoscentric.com/     The eCos and RedBoot
> experts
> 
> 
> -- 
> Before posting, please read the FAQ:
> http://ecos.sourceware.org/fom/ecos
> and search the list archive:
> http://ecos.sourceware.org/ml/ecos-discuss
> 
> 



	
		
__________________________________
Do you Yahoo!?
Yahoo! Mail - You care about security. So do we.
http://promotions.yahoo.com/new_mail

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] posix time functions

[David Brennan]

Here is my first attempt at a patch for this. I would have preferred 
putting the offset inside the Cyg_Clock class or the Cyg_RealTimeClock 
class, that way any "implementation" of gettime/settime could use that. 
However, I assumed you would not want the baggage in Cyg_Clock (even if 
it is CDL'able, it would end up in every instantiation of Cyg_Clock when 
configured enabled.) And there doesn't appear to be a "public" interface 
to the Cyg_RealTimeClock class that posix could call.

I was also wavering if the new term needed to be protected by a 
semaphore. I decided against it. (Due to the fact that an invalid 
gettime would have relatively low risk of creating problems combined 
with relative number of times that the time will be set within an 
application.) However, if you feel it is necessary, I will add one.

Once this patch gets finalized, the same thing should be done to the 
uitron interface. I'd be willing to post the code, but I haven't go the 
slightest idea how to test it.

Are there any other interfaces that can set/get the clock as "calendar 
time"? I could not find any.

David Brennan

Patch below:
Nick Garnett wrote:

>David Brennan <eCos@brennanhome.com> writes:
>
>  
>
>>We discovered a (perhaps) un-documented feature in the eCos posix
>>clock_settime and clock_gettime functions. Setting the time, changes
>>the system clock (number of ticks). The system clock should only be
>>changeable by the interrupt routine which is incrementing it. All
>>other code should really only have read access.
>>    
>>
>
>Hmm, we have also had complaints that clock_settime() doesn't change
>enough and want it to alter the wallclock time as well. I guess you
>cannot please everyone.
>
>  
>
>>Is there a chance this
>>could be changed? That is, if we fix it in the kernel, would that type
>>of patch be accepted.
>>    
>>
>
>A patch that maintains an offet for CLOCK_REALTIME from the main
>timer and used that when calculating timer timeouts would be
>fine. Depending on how much gets changed, we might need an assignment
>for it, though.
>
>  
>
>>I would think that setting the clock and
>>possibly losing alarms would be something that other people have
>>noticed before.
>>    
>>
>
>To be honest, I don't think that many people call clock_settime().
>It's mainly there for completeness.
>
>  
>
Index: compat/posix/current/ChangeLog
===================================================================
RCS file: /cvs/ecos/ecos/packages/compat/posix/current/ChangeLog,v
retrieving revision 1.48
diff -u -5 -w -r1.48 ChangeLog
--- compat/posix/current/ChangeLog    4 Oct 2004 11:49:19 -0000    1.48
+++ compat/posix/current/ChangeLog    22 Oct 2004 04:02:19 -0000
@@ -1,5 +1,12 @@
+2004-10-01  David Brennan  <eCos@brennanhome.com>
+
+    * cdl/posix.cdl:
+    * tests/timecheck.cxx:
+    * src/time.cxx:  Add capability to keep the real-time clock's ticks
+    monotonically increasing even through clock_settime.
+
 2004-10-01  Oyvind Harboe  <oyvind.harboe@zylin.com>
 
     * src/signal.cxx:  place the CYGBLD_ATTRIB_INIT_PRI such that it
     compiles for gcc 3.4.2 which is more picky about its placement.
 
Index: compat/posix/current/cdl/posix.cdl
===================================================================
RCS file: /cvs/ecos/ecos/packages/compat/posix/current/cdl/posix.cdl,v
retrieving revision 1.11
diff -u -5 -w -r1.11 posix.cdl
--- compat/posix/current/cdl/posix.cdl    24 Feb 2003 14:08:21 -0000    1.11
+++ compat/posix/current/cdl/posix.cdl    22 Oct 2004 04:02:19 -0000
@@ -150,10 +150,20 @@
         compile          time.cxx
     description      "This component provides configuration controls for
                       the POSIX clocks."
     }
 
+    cdl_option CYGCFG_POSIX_USE_RELATIVE_REALTIME {
+        display          "Keep real-time clock as relative time"
+        flavor           bool
+        default_value    0
+        description      "
+            This option ensures that the real-time clock will not go
+            backwards if using clock_settime to change the time. Using
+            other access functions can still cause the clock to change."
+    }
+
     # ----------------------------------------------------------------
     # Timers component
 
     cdl_option CYGPKG_POSIX_TIMERS {
     display          "POSIX timers"
@@ -302,10 +312,11 @@
             flavor  data
             no_define
             calculated {
                 "tests/pthread1 tests/pthread2 tests/pthread3 
tests/mutex3 tests/mqueue2"
                 . ((CYGPKG_POSIX_SIGNALS) ? " tests/mqueue1 
tests/signal1 tests/signal2 tests/signal3 tests/sigsetjmp tests/timer1 
tests/tm_basic" : "")
+                . ((CYGCFG_POSIX_USE_RELATIVE_REALTIME) ? " 
tests/timecheck" : "")
             }
             description   "
                 This option specifies the set of tests for the POSIX 
package."
         }
 
Index: compat/posix/current/src/time.cxx
===================================================================
RCS file: /cvs/ecos/ecos/packages/compat/posix/current/src/time.cxx,v
retrieving revision 1.9
diff -u -5 -w -r1.9 time.cxx
--- compat/posix/current/src/time.cxx    31 Jan 2003 11:53:14 -0000    1.9
+++ compat/posix/current/src/time.cxx    22 Oct 2004 04:02:20 -0000
@@ -127,10 +127,13 @@
 #define TIMER_ID_COOKIE_MASK (TIMER_ID_COOKIE_INC-1)
 static timer_t timer_id_cookie = TIMER_ID_COOKIE_INC;
 
 #endif // ifdef CYGPKG_POSIX_TIMERS
 
+#ifdef CYGCFG_POSIX_USE_RELATIVE_REALTIME
+static cyg_tick_count real_time_offset = 0;
+#endif
 //-----------------------------------------------------------------------------
 // new operator to allow us to invoke the constructor on
 // posix_timer.alarm_obj.
 
 inline void *operator new(size_t size,  cyg_uint8 *ptr) { return (void 
*)ptr; };
@@ -327,11 +330,15 @@
     if( !valid_timespec( tp ) )
         TIME_RETURN(EINVAL);
        
     cyg_tick_count ticks = cyg_timespec_to_ticks( tp );
 
+#ifdef CYGCFG_POSIX_USE_RELATIVE_REALTIME
+    real_time_offset = ticks - Cyg_Clock::real_time_clock->current_value();
+#else
     Cyg_Clock::real_time_clock->set_value( ticks );
+#endif
    
     TIME_RETURN(0);
 }
 
 //-----------------------------------------------------------------------------
@@ -347,10 +354,14 @@
     if( tp == NULL )
         TIME_RETURN(EINVAL);
    
     cyg_tick_count ticks = Cyg_Clock::real_time_clock->current_value();
 
+#ifdef CYGCFG_POSIX_USE_RELATIVE_REALTIME
+    ticks += real_time_offset;
+#endif
+
     cyg_ticks_to_timespec( ticks, tp );
 
     TIME_RETURN(0);
 }
   
--- /dev/null    2004-10-21 21:04:44.477000000 -0700
+++ compat/posix/current/tests/timecheck.cxx    2004-10-21 
17:51:20.500094400 -0700
@@ -0,0 +1,177 @@
+//==========================================================================
+//
+//        timecheck.cxx
+//
+//        Check that setting the time of day acts appropriately
+//
+//==========================================================================
+//####ECOSGPLCOPYRIGHTBEGIN####
+// -------------------------------------------
+// This file is part of eCos, the Embedded Configurable Operating System.
+// Copyright (C) 1998, 1999, 2000, 2001, 2002 Red Hat, Inc.
+// Copyright (C) 2003 Gary Thomas
+//
+// eCos is free software; you can redistribute it and/or modify it under
+// the terms of the GNU General Public License as published by the Free
+// Software Foundation; either version 2 or (at your option) any later 
version.
+//
+// eCos is distributed in the hope that it will be useful, but WITHOUT ANY
+// WARRANTY; without even the implied warranty of MERCHANTABILITY or
+// FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+// for more details.
+//
+// You should have received a copy of the GNU General Public License along
+// with eCos; if not, write to the Free Software Foundation, Inc.,
+// 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA.
+//
+// As a special exception, if other files instantiate templates or use 
macros
+// or inline functions from this file, or you compile this file and link it
+// with other works to produce a work based on this file, this file 
does not
+// by itself cause the resulting work to be covered by the GNU General 
Public
+// License. However the source code for this file must still be made 
available
+// in accordance with section (3) of the GNU General Public License.
+//
+// This exception does not invalidate any other reasons why a work based on
+// this file might be covered by the GNU General Public License.
+//
+// -------------------------------------------
+//####ECOSGPLCOPYRIGHTEND####
+//==========================================================================
+//#####DESCRIPTIONBEGIN####
+//
+// Author(s):     dmb
+// Contributors:  dmb
+// Date:          2004-10-20
+// Description:   Tests the Kernel Real Time Clock for correct operation
+//                while altering the time of day. Basically the test sets
+//                an alarm to go off 1 second in the future. Then the 
clock is
+//                set backwards by 9 seconds. The alarm should still go 
off in
+//                1 second, giving an apparent delta time of -8 
seconds. If the
+//                alarm fires in 10 seconds, then the clock_settime 
function
+//                messed with the real time clock. The behavior is 
controllable
+//                with the CYGCFG_POSIX_USE_RELATIVE_REALTIME. This 
test will
+//                fail with the option turned off, (but will not be built).
+//
+//####DESCRIPTIONEND####
+
+
+#include <pkgconf/kernel.h>
+
+#include <time.h>
+
+#include <cyg/infra/testcase.h>
+
+#include <cyg/kernel/clock.hxx>
+#include <cyg/kernel/clock.inl>
+#include <cyg/kernel/kapi.h>
+#include <cyg/kernel/sema.hxx>
+#include <cyg/kernel/thread.hxx>
+#include <cyg/kernel/thread.inl>
+#include <cyg/infra/diag.h>
+
+#define ENDPOINT 10
+
+static cyg_alarm_fn alarmfunc;
+static void alarmfunc( Cyg_Alarm *alarm, CYG_ADDRWORD data )
+{
+    Cyg_Binary_Semaphore *sp = (Cyg_Binary_Semaphore *)data;
+    sp->post();
+}
+
+
+static void entry0( CYG_ADDRWORD data )
+{
+    int i;
+    timespec last_time;
+    timespec this_time;
+    tm my_tm;
+    time_t my_timet;
+
+
+    // Set the time of day to something reasonable
+    my_tm.tm_sec = 0;
+    my_tm.tm_min = 0;
+    my_tm.tm_hour = 8;
+    my_tm.tm_mday = 20;
+    my_tm.tm_mon = 9;
+    my_tm.tm_year = 104;
+   
+    this_time.tv_sec = mktime( &my_tm );
+    this_time.tv_nsec = 0;
+   
+    clock_settime(CLOCK_REALTIME, &this_time);
+
+    // Set up variables for the alarm
+    Cyg_Clock *rtc = Cyg_Clock::real_time_clock;
+
+    Cyg_Binary_Semaphore sema;
+
+    Cyg_Alarm alarm( rtc, &alarmfunc, (CYG_ADDRWORD)&sema );
+
+    alarm.initialize( rtc->current_value() + 100, 100 );
+    alarm.enable();
+
+    // Get the current time of day
+    clock_gettime(CLOCK_REALTIME, &last_time);
+   
+    for ( i = 0; i < ENDPOINT; i++ ) {
+        // Set the time of day backwards by 9 seconds
+        this_time.tv_sec = last_time.tv_sec - 9;
+        this_time.tv_nsec = last_time.tv_nsec;
+        clock_settime(CLOCK_REALTIME, &this_time);
+
+        sema.wait();
+        // Get the current time of day
+        clock_gettime(CLOCK_REALTIME, &this_time);
+       
+        // Verify it is less than the previous time of day
+        CYG_TEST_CHECK(difftime(this_time.tv_sec,last_time.tv_sec) < 0.0,
+                       "When the time was set backwards, the real-time "
+                       "went backwards too.");
+
+        // Save the last time that we got
+        last_time = this_time;
+       
+        // print the current time of day
+        my_timet = this_time.tv_sec;
+        localtime_r( &my_timet, &my_tm );
+       
+        diag_printf( "INFO<current time %s -- RTC Low is %8d>\n",
+                     asctime(&my_tm), rtc->current_value_lo() );
+    }
+
+    CYG_TEST_PASS_FINISH("Clock set check OK");
+}
+
+static char stack[CYGNUM_HAL_STACK_SIZE_TYPICAL];
+static cyg_handle_t thread_handle;
+static cyg_thread thread;
+
+
+void clocktruth_main( void )
+{
+    CYG_TEST_INIT();
+    cyg_thread_create(3,                // Priority - just a number
+                      entry0,           // entry
+                      0,                // index
+                      0,                // no name
+                      &stack[0],        // Stack
+                      sizeof(stack),    // Size
+                      &thread_handle,   // Handle
+                      &thread           // Thread data structure
+        );
+    cyg_thread_resume(thread_handle);
+
+    Cyg_Scheduler::start();
+}
+
+externC void
+cyg_start( void )
+{
+#ifdef CYGSEM_HAL_STOP_CONSTRUCTORS_ON_FLAG
+    cyg_hal_invoke_constructors();
+#endif
+    clocktruth_main();
+}
+
+// EOF timecheck.cxx


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] posix time functions

[Nick Garnett]

David Brennan <david@brennanhome.com> writes:

> Here is my first attempt at a patch for this. I would have preferred
> putting the offset inside the Cyg_Clock class or the Cyg_RealTimeClock
> class, that way any "implementation" of gettime/settime could use
> that. However, I assumed you would not want the baggage in Cyg_Clock
> (even if it is CDL'able, it would end up in every instantiation of
> Cyg_Clock when configured enabled.) And there doesn't appear to be a
> "public" interface to the Cyg_RealTimeClock class that posix could
> call.

The POSIX layer is the correct place to do this, otherwise the
original problem of changing the time base under the feet of non-POSIX
timers would reoccur.

> 
> I was also wavering if the new term needed to be protected by a
> semaphore. I decided against it. (Due to the fact that an invalid
> gettime would have relatively low risk of creating problems combined
> with relative number of times that the time will be set within an
> application.) However, if you feel it is necessary, I will add one.
> 

This is good as far as it goes. However, you still need to make some
changes in timer_settime() and timer_gettime(). These need to add or
subtract the real_time_offset as appropriate. Consider the following
code for example:

    clock_gettime(CLOCK_REALTIME, &ts.it_value);
    ts.it_value.tv_sec += 5;
    timer_settime( timer, TIMER_ABSTIME, &ts, NULL );


I suspect that some changes will also need to be made in
pthread_cond_timedwait() and similar functions, so that the following
would work:

    clock_gettime(CLOCK_REALTIME, &ts);
    ts.tv_sec += 5;
    rc = 0;
    while (! mypredicate(&t) && rc == 0)
        rc = pthread_cond_timedwait(&cond, &mutex, &ts);


To avoid planting ifdefs all over the code I think that the offset
adjustment should be done in macros. These can be defined to do
nothing in the case where the option is disabled.

Strictly, each time the clock is changed, all timers, delays and
timeouts based upon it should be re-evaluated. However, this is
complicated and we don't currently preserve all the information
necessary to do this. 


The patch seems to have been corrupted a little, sending it as an
attachment should solve this.

> Once this patch gets finalized, the same thing should be done to the
> uitron interface. I'd be willing to post the code, but I haven't go
> the slightest idea how to test it.

I wouldn't worry too much about uitron at present. I'm not sure there
are many users.

-- 
Nick Garnett                    eCos Kernel Architect
http://www.ecoscentric.com/     The eCos and RedBoot experts


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] posix time functions

[David Brennan]

Nick Garnett wrote:

>David Brennan <david@brennanhome.com> writes:
>
>  
>
>>I was also wavering if the new term needed to be protected by a
>>semaphore. I decided against it. (Due to the fact that an invalid
>>gettime would have relatively low risk of creating problems combined
>>with relative number of times that the time will be set within an
>>application.) However, if you feel it is necessary, I will add one.
>>
>>    
>>
>
>This is good as far as it goes. However, you still need to make some
>changes in timer_settime() and timer_gettime(). These need to add or
>subtract the real_time_offset as appropriate. Consider the following
>code for example:
>
>    clock_gettime(CLOCK_REALTIME, &ts.it_value);
>    ts.it_value.tv_sec += 5;
>    timer_settime( timer, TIMER_ABSTIME, &ts, NULL );
>
>  
>
Hmmm, I did not think of people accessing it that way, I assumed that 
clock_gettime would only be used to get a calendar time. I had assumed 
(probably incorrectly) that to get the current real-time, one would use 
something like cyg_current_time(). Solving it like you suggest is only 
going to bring back the problem that I initially had. I haven't looked 
at the API closely, would it be safe to assume the code written above 
could be written using timer_gettime first?  The idea was that if you 
change the calendar time, you do not want all of the absolute timers to 
change their alarm times. In my application, I will set the time using a 
pseudo NTP type function about once an hour. Due to the timing 
resolution of the i386 platform with a base clock frequency of 1 kHz, I 
will set the time backwards about 200 mSec each hour.  This causes all 
of my alarms to get delayed by an extra 200 mSec. Therefore, I want to 
be able to change the time that posix thinks it is without affecting any 
"real" real-time timers, none of which are set from the clock_gettime 
function.
.

>I suspect that some changes will also need to be made in
>pthread_cond_timedwait() and similar functions, so that the following
>would work:
>
>    clock_gettime(CLOCK_REALTIME, &ts);
>    ts.tv_sec += 5;
>    rc = 0;
>    while (! mypredicate(&t) && rc == 0)
>        rc = pthread_cond_timedwait(&cond, &mutex, &ts);
>
>
>To avoid planting ifdefs all over the code I think that the offset
>adjustment should be done in macros. These can be defined to do
>nothing in the case where the option is disabled.
>
>Strictly, each time the clock is changed, all timers, delays and
>timeouts based upon it should be re-evaluated. However, this is
>complicated and we don't currently preserve all the information
>necessary to do this. 
>
>  
>
I think my comment above applies to this also.

>The patch seems to have been corrupted a little, sending it as an
>attachment should solve this.
>
>  
>
I can resend it as an attachment when we get this figured out. (Is it 
readable?)

David Brennan

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

[ECOS] eCos x Cirrus EP9302

[sraposo]

Hi!

Does someone know if there is any eCos already ported for Cirrus EP9302?

Regards,

Sérgio


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] posix time functions

[Nick Garnett]

David Brennan <eCos@brennanhome.com> writes:

> Nick Garnett wrote:
> 
> >David Brennan <david@brennanhome.com> writes:
> >
> >
> >>I was also wavering if the new term needed to be protected by a
> >>semaphore. I decided against it. (Due to the fact that an invalid
> >>gettime would have relatively low risk of creating problems combined
> >>with relative number of times that the time will be set within an
> >>application.) However, if you feel it is necessary, I will add one.
> >>
> >>
> >
> >This is good as far as it goes. However, you still need to make some
> >changes in timer_settime() and timer_gettime(). These need to add or
> >subtract the real_time_offset as appropriate. Consider the following
> >code for example:
> >
> >    clock_gettime(CLOCK_REALTIME, &ts.it_value);
> >    ts.it_value.tv_sec += 5;
> >    timer_settime( timer, TIMER_ABSTIME, &ts, NULL );
> >
> >
> Hmmm, I did not think of people accessing it that way, I assumed that
> clock_gettime would only be used to get a calendar time. I had assumed
> (probably incorrectly) that to get the current real-time, one would
> use something like cyg_current_time().

No. clock_gettime() is the intended source of timestamps for
generating absolute times in all the POSIX timer functions. It's the
only thing that returns a struct timespec, other functions return
different things.

> Solving it like you suggest is
> only going to bring back the problem that I initially had.

I probably misunderstood what problem you were seeing. I was trying to
solve the problem of non-POSIX timeouts (in the netstack for example)
being affected by calls to clock_settime(). It looks like you were
more concerned over its effect on POSIX timeouts.

Strict POSIX conformance requires, when clock_settime() is called, for
all abosolute timeouts to track the change in the time base. Relative
timeouts should not change and remain ready to trigger after the same
interval.

> I haven't
> looked at the API closely, would it be safe to assume the code written
> above could be written using timer_gettime first?

That just gets the time until the expiry of an active timer, I don't
think it would make any difference.

> The idea was that
> if you change the calendar time, you do not want all of the absolute
> timers to change their alarm times. In my application, I will set the
> time using a pseudo NTP type function about once an hour. Due to the
> timing resolution of the i386 platform with a base clock frequency of
> 1 kHz, I will set the time backwards about 200 mSec each hour.  This
> causes all of my alarms to get delayed by an extra 200
> mSec. Therefore, I want to be able to change the time that posix
> thinks it is without affecting any "real" real-time timers, none of
> which are set from the clock_gettime function.

This seems a somewhat kludgy solution. Exactly where is the inaccuracy
coming from?

Using the PIT to keep track of real time over a long period is hard
since it runs at a very awkward frequency -- although I believe it is
reasonably accurate at a 200Hz interrupt rate. It is OK for handling
timing requirements of less that a minute or so.

> The more that I think about this, I'm wondering if the right solution
> is to add functions like cyg_gettimeofday and cyg_settimeofday. The
> only reason we were using the Posix function (and in fact, I think the
> only reason we are including the POSIX package), is because it is the
> only public function to adjust the time of day. However, I can see
> where it would lead to confusion. Additionally, we could make these
> functions conditionally (with CDL) set the wall-clock, as some people
> had previously requested.

Does your i386 platform have the standard Dallas RTC device? If so
then you could use that to track real time more accurately. The Dallas
device also provides a 1KHz interrupt which is more accurate than the
PIT, and is synchronized to the real time clock. It may be easier to
add an option to switch the eCos timer over to using that.


-- 
Nick Garnett                    eCos Kernel Architect
http://www.ecoscentric.com/     The eCos and RedBoot experts


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] posix time functions

[David Brennan]

Nick Garnett wrote:

>David Brennan <eCos@brennanhome.com> writes:
>
>  
>
>>Nick Garnett wrote:
>>
>>    
>>
>>>David Brennan <david@brennanhome.com> writes:
>>>
>>>
>>>      
>>>
>>>>I was also wavering if the new term needed to be protected by a
>>>>semaphore. I decided against it. (Due to the fact that an invalid
>>>>gettime would have relatively low risk of creating problems combined
>>>>with relative number of times that the time will be set within an
>>>>application.) However, if you feel it is necessary, I will add one.
>>>>
>>>>
>>>>        
>>>>
>>>This is good as far as it goes. However, you still need to make some
>>>changes in timer_settime() and timer_gettime(). These need to add or
>>>subtract the real_time_offset as appropriate. Consider the following
>>>code for example:
>>>
>>>   clock_gettime(CLOCK_REALTIME, &ts.it_value);
>>>   ts.it_value.tv_sec += 5;
>>>   timer_settime( timer, TIMER_ABSTIME, &ts, NULL );
>>>
>>>
>>>      
>>>
>>Hmmm, I did not think of people accessing it that way, I assumed that
>>clock_gettime would only be used to get a calendar time. I had assumed
>>(probably incorrectly) that to get the current real-time, one would
>>use something like cyg_current_time().
>>    
>>
>
>No. clock_gettime() is the intended source of timestamps for
>generating absolute times in all the POSIX timer functions. It's the
>only thing that returns a struct timespec, other functions return
>different things.
>
>  
>
>>Solving it like you suggest is
>>only going to bring back the problem that I initially had.
>>    
>>
>
>I probably misunderstood what problem you were seeing. I was trying to
>solve the problem of non-POSIX timeouts (in the netstack for example)
>being affected by calls to clock_settime(). It looks like you were
>more concerned over its effect on POSIX timeouts.
>
>Strict POSIX conformance requires, when clock_settime() is called, for
>all abosolute timeouts to track the change in the time base. Relative
>timeouts should not change and remain ready to trigger after the same
>interval.
>
>  
>
You are correct, my particular problem is setting the clock with the 
POSIX function, then causing issues with non-POSIX timers. But I was 
thinking that it would make sense for the same change to apply to the 
POSIX timers also. That was when I checked the POSIX specifications and 
decided that perhaps this isn't the right way to fix this.

>>The idea was that
>>if you change the calendar time, you do not want all of the absolute
>>timers to change their alarm times. In my application, I will set the
>>time using a pseudo NTP type function about once an hour. Due to the
>>timing resolution of the i386 platform with a base clock frequency of
>>1 kHz, I will set the time backwards about 200 mSec each hour.  This
>>causes all of my alarms to get delayed by an extra 200
>>mSec. Therefore, I want to be able to change the time that posix
>>thinks it is without affecting any "real" real-time timers, none of
>>which are set from the clock_gettime function.
>>    
>>
>
>This seems a somewhat kludgy solution. Exactly where is the inaccuracy
>coming from?
>
>Using the PIT to keep track of real time over a long period is hard
>since it runs at a very awkward frequency -- although I believe it is
>reasonably accurate at a 200Hz interrupt rate. It is OK for handling
>timing requirements of less that a minute or so.
>  
>
Yes the problem is with that awkward frequency. I don't have the math 
handy, but we calculated that if we set the PIT to an interrupt rate of 
1kHz, we would end up off by about 180 mSec per hour. Our system 
actually has multiple computers which generate messages with "real-time" 
time stamps which are often compared with each other for diagnostic 
purposes. So we need mSec accuracy on those messages.

>  
>
>>The more that I think about this, I'm wondering if the right solution
>>is to add functions like cyg_gettimeofday and cyg_settimeofday. The
>>only reason we were using the Posix function (and in fact, I think the
>>only reason we are including the POSIX package), is because it is the
>>only public function to adjust the time of day. However, I can see
>>where it would lead to confusion. Additionally, we could make these
>>functions conditionally (with CDL) set the wall-clock, as some people
>>had previously requested.
>>    
>>
>
>Does your i386 platform have the standard Dallas RTC device? If so
>then you could use that to track real time more accurately. The Dallas
>device also provides a 1KHz interrupt which is more accurate than the
>PIT, and is synchronized to the real time clock. It may be easier to
>add an option to switch the eCos timer over to using that.
>
>
>  
>
Really? I did not think the Dallas chip could interrupt that quickly. (I 
thought it was on the order of 1 Hz). However, I will still need to be 
able to set the clock without affecting any of the OS timers.

David

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

Re: [ECOS] posix time functions

[Nick Garnett]

David Brennan <eCos@brennanhome.com> writes:

> Nick Garnett wrote:
> >Using the PIT to keep track of real time over a long period is hard
> >since it runs at a very awkward frequency -- although I believe it is
> >reasonably accurate at a 200Hz interrupt rate. It is OK for handling
> >timing requirements of less that a minute or so.
> >
> Yes the problem is with that awkward frequency. I don't have the math
> handy, but we calculated that if we set the PIT to an interrupt rate
> of 1kHz, we would end up off by about 180 mSec per hour. Our system
> actually has multiple computers which generate messages with
> "real-time" time stamps which are often compared with each other for
> diagnostic purposes. So we need mSec accuracy on those messages.

Of course this is the sort of thing that the numerator/denominator
representation of the clock resolution is meant to deal with. The idea
is to adjust those values to more accurately represent the interrupt
frequency the hardware actually gives you, rather than the
approximation used to calculate the timer period. The result is that
eCos ticks end up slightly shorter than 1ms and you don't get the
drift.

> >Does your i386 platform have the standard Dallas RTC device? If so
> >then you could use that to track real time more accurately. The Dallas
> >device also provides a 1KHz interrupt which is more accurate than the
> >PIT, and is synchronized to the real time clock. It may be easier to
> >add an option to switch the eCos timer over to using that.
> >
> >
> >
> Really? I did not think the Dallas chip could interrupt that
> quickly. (I thought it was on the order of 1 Hz).

Specifically it generates a 1024Hz interrupt. However, since there is
no underying fast counter, you cannot do timing more accurately than
that, for hal_delay_us() for example, so it is not a good choice for
the main eCos timer.

> However, I will
> still need to be able to set the clock without affecting any of the OS
> timers.

If you can sort out the timer frequency, then you should only really
need to set it once at startup, before any timeouts are started. If
you use the RTC then you may not need to set it at all.


-- 
Nick Garnett                    eCos Kernel Architect
http://www.ecoscentric.com/     The eCos and RedBoot experts


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

[ECOS] clock_settime() function

[Sam Pham]

Hi all,

I was wondering what the context of calling function
clock_settime() should be.  Is it ISR, DSR, or thread?
 I call this function inside my ISR, but when I call
clock_gettime() after that (with 1 second delay), I
don't see the time was set accordingly.  

Thanks.


		
__________________________________
Do you Yahoo!?
Yahoo! Mail Address AutoComplete - You start. We finish.
http://promotions.yahoo.com/new_mail 

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

[ECOS] clock_settime() help

[Sam Pham]

Hi all,

I'm having trouble with calling the clock_settime()
function.  My application has the need of setting the
clock to the current time.  At first, I called
clock_settime() inside my ISR, and then outside of
this ISR, I called clock_gettime() to verify if the
current time was set.  I noticed that sometimes
clock_settime() failed because clock_gettime()
returned a non-current time value.  So I decided to
call clock_settime() in a thread with all interrupts
disabled by calling function cyg_interrupt_disable(). 
However, I still observed the same failure as before.

Could anyone explain to me how I should be using
function clock_settime()?  Its implementation is very
simple - just change the counter value.  However, it
looks like if the kernel is in certain states, then it
won't allow anything to change this counter value.

Thanks.


		
__________________________________
Do you Yahoo!?
Yahoo! Mail Address AutoComplete - You start. We finish.
http://promotions.yahoo.com/new_mail 

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

[ECOS] tcsendbreak

[Sam Pham]

Hi all,

The function tcsendbreak() in file
io/serial/current/src/common/termios.c isn't
implemented and simply returns -EINVAL.  Does the
latest ecos version implement it?  If not, any
suggestions of how to go about implementing it?

Thanks,

Sam


		
__________________________________ 
Start your day with Yahoo! - Make it your home page! 
http://www.yahoo.com/r/hs

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss