Hello L4 hackers,
we've encountered a problem with an IPC transmission based on DICE stubs generated from the following interface definition:
UInt32 process( [in, size_is(bufferALen), max_is(MAX_BUFFER_SIZE)] UInt8* bufferA, [in] UInt32 bufferALen, [in, size_is(bufferBLen), max_is(MAX_BUFFER_SIZE)] UInt8* bufferB , [in] UInt32 bufferBLen, [out, size_is(*bufferCLen), max_is(MAX_BUFFER_SIZE)] UInt8** bufferC, [out] UInt32* bufferCLen);
UInt8 is defined as unsigned char, UInt32 is defined as unsigned long, MAX_BUFFER_SIZE is 1500.
We are using dice version 3.1.0 with the command line options -fforce-corba-alloc -nostdinc -P-DL4API_l4v2 and the latest OTC snapshot.
In this configuration bufferB is not transmitted correctly but is received as corrupt data on the server side (which is an L4Linux application).
Is there anything wrong with our interface definition? Any hints appreciated!
Michael
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Michael Scheibel wrote on 09.02.2007 11:52 this:
Hello L4 hackers,
we've encountered a problem with an IPC transmission based on DICE stubs generated from the following interface definition:
UInt32 process( [in, size_is(bufferALen), max_is(MAX_BUFFER_SIZE)] UInt8* bufferA, [in] UInt32 bufferALen, [in, size_is(bufferBLen), max_is(MAX_BUFFER_SIZE)] UInt8* bufferB , [in] UInt32 bufferBLen, [out, size_is(*bufferCLen), max_is(MAX_BUFFER_SIZE)] UInt8** bufferC, [out] UInt32* bufferCLen);
UInt8 is defined as unsigned char, UInt32 is defined as unsigned long, MAX_BUFFER_SIZE is 1500.
We are using dice version 3.1.0 with the command line options -fforce-corba-alloc -nostdinc -P-DL4API_l4v2 and the latest OTC snapshot.
In this configuration bufferB is not transmitted correctly but is received as corrupt data on the server side (which is an L4Linux application).
Is there anything wrong with our interface definition? Any hints appreciated!
Sorry, for the delay. Confirmed as a Dice bug. I did not get into fixing this yet.
Greets, Ron. - -- Mit freundlichen Gruessen / with regards ra3 @ inf.tu-dresden.de http://os.inf.tu-dresden.de/~ra3/
Hi,
Ronald Aigner wrote on 12.02.2007 10:12 this:
Is there anything wrong with our interface definition? Any hints appreciated!
Sorry, for the delay. Confirmed as a Dice bug. I did not get into fixing this yet.
please find attached a diff for a work around.
HTH, Ron.
Ronald Aigner wrote on 14.02.2007 18:44 this:
please find attached a diff for a work around.
Sorry, wrong order. Now, please find attached a patch for a work around.
HTH, Ron.
Am Mittwoch 14 Februar 2007 19:34 schrieb Ronald Aigner:
Ronald Aigner wrote on 14.02.2007 18:44 this:
please find attached a diff for a work around.
Sorry, wrong order. Now, please find attached a patch for a work around.
HTH, Ron.
Hi Ron,
the patch seems to solve the problem, even for dice 3.1.0.
Thank you! Michael
l4-hackers@os.inf.tu-dresden.de