Crash related to FontCache

[humdinger]

This is hrev52827, 32bit.

[Please adjust the title of this ticket if necessary.)

Had an app_server crash after a lengthy Web+ session. This from the syslog (full debug report attached:

KERN: runtime_loader: /boot/system/lib/x86/libavformat.so.58.20.100: Could not map image: Out of memory
Last message repeated 1 time
KERN: acquire_advisory_lock(vnode = 0xdb07b128, flock = 0x80286efc, wait = no)
KERN: Last message repeated 2 times.
KERN: Last message repeated 3 times.
KERN: slab memory manager: created area 0xdc801000 (27441)
KERN: acquire_advisory_lock(vnode = 0xdb07b128, flock = 0x80286efc, wait = no)
KERN: Last message repeated 2 times.
KERN: Last message repeated 3 times.
Last message repeated 1 time
KERN: Last message repeated 6 times.
KERN: runtime_loader: /boot/system/lib/x86/libIlmImf-2_2.so.23.0.0: Could not map image: Out of memory
KERN: 2459: DEBUGGER: _numBlocks > 0
KERN: debug_server: Thread 2459 entered the debugger: Debugger call: `_numBlocks > 0'
KERN: stack trace, current PC 0x625e5114  commpage_syscall + 0x4:
KERN:   (0x712f98e8)  0x141ecc2  __assert_fail + 0x6e
KERN:   (0x712f9928)  0x14967ce  free__Q28BPrivate11processHeapPv + 0x8a
KERN:   (0x712f9968)  0x149832e  free + 0xae
KERN:   (0x712f9998)  0x140b2e0  __builtin_delete + 0x24
KERN:   (0x712f99c8)  0xa9a2f6  _._14FontCacheEntry + 0xb6
KERN:   (0x712f9a08)  0x7a5c8a  LastReferenceReleased__14BReferenceable + 0x2e
KERN:   (0x712f9a38)  0x7a5c47  ReleaseReference__14BReferenceable + 0x3f
KERN:   (0x712f9a68)  0xa994c8  _ConstrainEntryCount__9FontCache + 0x640
KERN:   (0x712f9b28)  0xa98b3d  FontCacheEntryFor__9FontCacheRC10ServerFontb + 0x1b9
KERN:   (0x712f9d78)  0xa65526  GetBoundingBoxes__10ServerFontPCcllP5BRectb16font_metric_modeG16escapement_deltaT5 + 0x43e
KERN:   (0x712fa1a8)  0xa5c692  _DispatchMessage__9ServerApplRQ28BPrivate12LinkReceiver + 0x54f2
KERN:   (0x712fb248)  0xa5f504  _MessageLooper__9ServerApp + 0x1b0
KERN:   (0x712fb308)  0xa4ebec  _message_thread__13MessageLooperPv + 0x28
KERN:   (0x712fb338)  0x1415039  thread_entry + 0x29
KERN: 2459: DEBUGGER: abort() called
KERN: <BEEP>
KERN: vesa: a

[waddlesplash]

Still reproducible?

[humdinger]

I don't think it ever was actually "reproducible", which is the problem with debugging most app_server crashes, I suppose... Haven't seen it again, though, I think.

[waddlesplash]

It's quite possible the reference-counting changes in 2020 fixed this, anyway.