QuickLaunch crashes on quit with new allocator

[diversys]

Debug information for team /boot/system/apps/QuickLaunch/QuickLaunch (2308):
CPU(s): 2x Intel Core™ i7-3635QM
Memory: 2.00 GiB total, 405.51 MiB used
Haiku revision: hrev53137 May 16 2019 19:57:09 (x86_64)

Active Threads:
	thread 2318: team 2308 debug task 
	thread 2308: QuickLaunch (main)
		state: Call (thread 2308 tried accessing address 0x57ae40 which is not allocated (base: 0x57ae40, size: 440, alignment: 16, allocated by thread: 2308, freed by thread: 2317))

		Frame		IP			Function Name
		-----------------------------------------------
		00000000	0x1645e68db7f	_kern_debugger + 0x7 
			Disassembly:
				_kern_debugger:
				0x000001645e68db78:   48c7c0e1000000  mov $0xe1, %rax
				0x000001645e68db7f:             0f05  syscall  <--

		0x7f7719d868d0	0x1645e736130	panic(char const*, ...) + 0xc0 
		0x7f7719d86930	0x1645e7365ac	guarded_heap_segfault_handler(int, __siginfo_t*, void*) + 0x1ac 
		0x7f7719d86930	0x7ffaa33c123b	commpage_signal_handler + 0x2b 
		0x7f7719d86d20	0x1c075b0b0e1	QLSettings::~QLSettings() + 0x21 
		0x7f7719d86d90	0x1c075b0b827	QLApp::~QLApp() + 0x97 
		0x7f7719d86fb0	0x1c075b054b8	main + 0x28 
		0x7f7719d86fe0	0x1c075b055fc	_start + 0x4c 
		0x7f7719d87010	0x4e524f96c0	runtime_loader + 0x100 
		00000000	0x7ffaa33c1260	commpage_thread_exit + 0 

[humdingerb]

I had a quick look, and moved saving the search term from the app's d'tor to the MainWindow d'tor (c829080). That didn't solve the issue though... :}

Interesting: there's no issue on 32bit...

[humdingerb]

Seems to be fixed after commenting out https://github.com/humdingerb/quicklaunch/blob/master/QLSettings.cpp#L97

Not sure exactly why...

[waddlesplash]

The issue always appears with the guarded heap, which isn't new. The new default allocator just has memory usage patterns enough like the guarded heap to trigger this.

Not sure why it's not an issue on 32-bit, though.

[humdingerb]

Thanks for the hints, waddlesplash!

@diversys : Should be fixed now. Can you retest? I'll push a bugfix release if it's OK later...