[Ns-bugs] [Bug 486] Patches for compilation with icc

Wed Feb 4 12:32:02 PST 2009

http://www.nsnam.org/bugzilla/show_bug.cgi?id=486

--- Comment #3 from TimoB <timo.bingmann at student.kit.edu>  2009-02-04 15:32:02 EDT ---
Yes, sorry, some are warnings some are errors.

(In reply to comment #2)
> (From update of attachment 368 [details])
> Eh ? The following change can't possibly be right: what is the error message
> output by icc ?

This is about the default arguments, they should only be declared in the
function declaration.

../src/core/object.h(406): warning #845: specifying a default argument when
redeclaring an unreferenced function template is nonstandard
  CreateObject (std::string n1 = "", const AttributeValue & v1 =
EmptyAttributeValue (),
  ^

../src/core/object.h(406): warning #350: redefinition of default argument
  CreateObject (std::string n1 = "", const AttributeValue & v1 =
EmptyAttributeValue (),
  ^

> >diff -r 3e819441bc75 -r 0ad61c84f3c6 src/core/object.h
> >--- a/src/core/object.h	Tue Feb 03 06:56:47 2009 -0800
> >+++ b/src/core/object.h	Wed Feb 04 13:08:04 2009 +0100
> >@@ -403,15 +403,15 @@ Ptr<T> CreateObject (const AttributeList
> > 
> > template <typename T>
> > Ptr<T> 
> >-CreateObject (std::string n1 = "", const AttributeValue & v1 = EmptyAttributeValue (),
> >-              std::string n2 = "", const AttributeValue & v2 = EmptyAttributeValue (),
> >-              std::string n3 = "", const AttributeValue & v3 = EmptyAttributeValue (),
> >-              std::string n4 = "", const AttributeValue & v4 = EmptyAttributeValue (),
> >-              std::string n5 = "", const AttributeValue & v5 = EmptyAttributeValue (),
> >-              std::string n6 = "", const AttributeValue & v6 = EmptyAttributeValue (),
> >-              std::string n7 = "", const AttributeValue & v7 = EmptyAttributeValue (),
> >-              std::string n8 = "", const AttributeValue & v8 = EmptyAttributeValue (),
> >-              std::string n9 = "", const AttributeValue & v9 = EmptyAttributeValue ())
> >+CreateObject (std::string n1 , const AttributeValue & v1,
> >+              std::string n2 , const AttributeValue & v2,
> >+              std::string n3 , const AttributeValue & v3,
> >+              std::string n4 , const AttributeValue & v4,
> >+              std::string n5 , const AttributeValue & v5,
> >+              std::string n6 , const AttributeValue & v6,
> >+              std::string n7 , const AttributeValue & v7,
> >+              std::string n8 , const AttributeValue & v8,
> >+              std::string n9 , const AttributeValue & v9)
> > {
> >   AttributeList attributes;
> >   if (n1 == "")
> 
> What warning are you fixing below ?

It's an error, which I cannot tell you why it occurs:

../src/internet-stack/ipv4-l3-protocol.cc(1071): error: cast to type
"ns3::Ipv4Address ()" is not allowed
    if ((interface->GetAddress ()) != (Ipv4Address ())
                                       ^

../src/internet-stack/ipv4-l3-protocol.cc(1072): error: expected an expression
        && (interface->GetNetworkMask ()) != (Ipv4Mask ()))
        ^

> >diff -r 3e819441bc75 -r 0ad61c84f3c6 src/internet-stack/ipv4-l3-protocol.cc
> >--- a/src/internet-stack/ipv4-l3-protocol.cc	Tue Feb 03 06:56:47 2009 -0800
> >+++ b/src/internet-stack/ipv4-l3-protocol.cc	Wed Feb 04 13:08:04 2009 +0100
> >@@ -1068,8 +1068,8 @@ Ipv4L3Protocol::SetUp (uint32_t i)
> >   // If interface address and network mask have been set, add a route
> >   // to the network of the interface (like e.g. ifconfig does on a
> >   // Linux box)
> >-  if ((interface->GetAddress ()) != (Ipv4Address ())
> >-      && (interface->GetNetworkMask ()) != (Ipv4Mask ()))
> >+  if ( ! interface->GetAddress ().IsEqual (Ipv4Address ())
> >+       && (interface->GetNetworkMask ()) != (Ipv4Mask ()))
> >     {
> >       AddNetworkRouteTo (interface->GetAddress ().CombineMask (interface->GetNetworkMask ()),
> >                          interface->GetNetworkMask (), i);
> 
> This is not really a warning fix, right ?

Warning about comparision of unsigned int >= 0. I changed the code to actually
compare something.

> >diff -r 3e819441bc75 -r 0ad61c84f3c6 src/node/queue.cc
> >--- a/src/node/queue.cc	Tue Feb 03 06:56:47 2009 -0800
> >+++ b/src/node/queue.cc	Wed Feb 04 13:08:04 2009 +0100
> >@@ -84,12 +84,12 @@ Queue::Dequeue (void)
> > 
> >   if (packet != 0)
> >     {
> >+      NS_ASSERT (m_nBytes >= packet->GetSize ());
> >+      NS_ASSERT (m_nPackets > 0);
> >+
> >       m_nBytes -= packet->GetSize ();
> >       m_nPackets--;
> > 
> >-      NS_ASSERT (m_nBytes >= 0);
> >-      NS_ASSERT (m_nPackets >= 0);
> >-
> >       NS_LOG_LOGIC("m_traceDequeue (packet)");
> > 
> >       m_traceDequeue (packet);
> 
> What warning are you fixing below ? LLU is not very portable so, adding:
> uint64_t pow10 (uint8_t n)
> {
>   NS_ASSERT (n > 0);
>   uint64_t result = 1;
>   for (uint8_t i = 0; i < n; i++)
>     {
>       result *= 10;
>     }
>   return result;
> }
> would be probably more portable.

Interesting to talk about portability between compilers here ;).

This is actually a runtime error. Any example program will segfault, because
these static consts are incorrectly initialized, gdb says they are 0 at
segfault time.

Your idea with pow10 did not work.

The issue is about initialization order: in my dump WifiMac::GetDefaultSifs()
is called during initialization _before_ the pow10() is calculated. SIFS is
returned with MicroSeconds(), which uses one of the factors.

static const uint64_t MS_FACTOR = (uint64_t)1000;
static const uint64_t US_FACTOR = (uint64_t)1000000;
static const uint64_t NS_FACTOR = (uint64_t)1000000 * (uint64_t)1000;
static const uint64_t PS_FACTOR = (uint64_t)1000000 * (uint64_t)1000000;
static const uint64_t FS_FACTOR = (uint64_t)1000000 * (uint64_t)1000000 *
(uint64_t)1000;

The above works. The compilers (gcc/icc) seem to be able to fold the
expression.

> >diff -r 3e819441bc75 -r 0ad61c84f3c6 src/simulator/time.cc
> >--- a/src/simulator/time.cc	Tue Feb 03 06:56:47 2009 -0800
> >+++ b/src/simulator/time.cc	Wed Feb 04 13:08:04 2009 +0100
> >@@ -32,11 +32,11 @@ namespace ns3 {
> > 
> > namespace TimeStepPrecision {
> > 
> >-static const uint64_t MS_FACTOR = (uint64_t)pow(10,3);
> >-static const uint64_t US_FACTOR = (uint64_t)pow(10,6);
> >-static const uint64_t NS_FACTOR = (uint64_t)pow(10,9);
> >-static const uint64_t PS_FACTOR = (uint64_t)pow(10,12);
> >-static const uint64_t FS_FACTOR = (uint64_t)pow(10,15);
> >+static const uint64_t MS_FACTOR = 1000LLU;
> >+static const uint64_t US_FACTOR = 1000000LLU;
> >+static const uint64_t NS_FACTOR = 1000000000LLU;
> >+static const uint64_t PS_FACTOR = 1000000000000LLU;
> >+static const uint64_t FS_FACTOR = 1000000000000000LLU;
> > static uint64_t g_tsPrecFactor = NS_FACTOR;
> > 
> > static GlobalValue g_precisionDefaultValue ("TimeStepPrecision", 
> 

-- 
Configure bugmail: http://www.nsnam.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.