Second shared region breaks Ipc_start

Sam Body

var SharedRegion = xdc.useModule('ti.sdo.ipc.SharedRegion');
	SharedRegion.numEntries = 2;
	SharedRegion.translate = false;
/*
	SharedRegion.setEntryMeta(1,
    	{
		base: 0x90000000,
		len:  0x70000000,
		ownerProcId: 0,
		isValid: true,
		cacheEnable: false,
		cacheLineSize: 128,
		createHeap: true,
		name: "external_shared_mem",
	    });
*/
	SharedRegion.setEntryMeta(0,
    	{
		base: 0x0C000000, 
		len:  0x00200000,
		ownerProcId: 0,
		isValid: true,
		cacheEnable: true,
		cacheLineSize: 128,
		createHeap: true,
		name: "internal_shared_mem",
	    });

var Ipc = xdc.useModule('ti.sdo.ipc.Ipc');
	Ipc.procSync = Ipc.ProcSync_ALL;

Uncommenting external_shared_mem leads to assertion when Ipc_start() is called

ti.sdo.ipc.heaps.HeapMemMP: line 335: assertion failure: A_internal: An internal error has occurred
xdc.runtime.Error.raise: terminating execution

What is wrong with configuration?

over 11 years ago

0 Sam Body over 11 years ago

Intellectual 370 points

You have a bug somewhere in IPC package, my friends.
With these settings no assertion is raised:
base: 0x90000000,
len: 0x70000000 - 4,
Obviously you have 32-bit pointer overflow.

0 Robert Tivy over 11 years ago in reply to Sam Body

TI__Mastermind 18260 points

Sam Body said:
Obviously you have 32-bit pointer overflow.

Thankyou for your report of a bug, it is indeed overflowing when checking an address against the end of the buffer (it checks for addr < (base + len) which fails when base + len = 2^32).

I will file a bug for this. At present there is an implied constraint that (base + len) < 0x100000000 (i.e., 2^32).

Thanks & Regards,

- Rob

Processors

Processors forum

Second shared region breaks Ipc_start