https://bugs.kde.org/show_bug.cgi?id=384442
--- Comment #4 from Julian Seward ---
There's a bunch of related hacks in the arm(32) and arm64 front
ends. Grep for "earlyWriteback" or some such. They are probably
related.
The fundamental problem though is that these insns simply aren't
representa
https://bugs.kde.org/show_bug.cgi?id=384442
--- Comment #3 from John Reiser ---
Here is the output from "--trace-flags=1000", condensed and commented:
=
(arm) 0x49BD90C: mov r12, r13
(arm) 0x49BD910: stmdb r13!, {0xDFF0}
(arm) 0x49BD914: sub r11, r12, #0x4
https://bugs.kde.org/show_bug.cgi?id=384442
--- Comment #2 from John Reiser ---
Context: [valgrind-users] by Kacper Kowalski on 2017-08-31 and replies through
the following week.
*
processor: 0
model name: ARMv7 Processor rev 10 (v7l)
BogoMIPS: 132.00
Features: half thumb fa
https://bugs.kde.org/show_bug.cgi?id=384442
--- Comment #1 from John Reiser ---
Both problems can be solved by a special case for 'ldmdb' (LoaD Multiple
registers, Decrementing the address Before each fetch). If either r15(==pc) or
r13(==sp) is to be loaded, then assign the new value to a tempor
