realloc from a module broken on Mesolithic systems

I’ve seen this a number of times:

Since realloc from a module is broken in the CLib built into RISC OS 3.10

What is the actual behaviour of this? In what way is it broken?
(doesn’t [m|c|re]alloc/free from a module use the RMA?)

Hi Rick
I use the ModMalloc library which states:

/*

For use by module apps which use malloc. malloc tries to use app memory

if called from user mode, so will always return 0 if the module app has

zero app space.

These malloc replacements use OS_Module 6/7 instead.
Due to problems with atexit() in module code, ModMalloc doesn’t do a

‘atexit( ModMalloc_FreeAll)’, which would ensure all allocated memory

was freed. Instead, your module finalisation code should call

ModMalloc_FreeAll explicitly.

 */

I don’t know if this is still true today?

The grey cell must be dimming, Rick! Put another shilling in. ;-)

https://www.riscosopen.org/forum/forums/4/topics/17355#posts-133691

The grey cell must be dimming

Uh-huh. First sign of… ;)

Thanks for the pointers.

First sign of… ;)

Age enhanced effects of herpes simplex (apparently/allegedly)