The physical memory access "bit31" mapping (0--0x30000000 => 0x80000000--0xB0000000) as well as privileged (kernel) mappings confuse debugger clients like IDA's which interpret the value of the registers.
Access to those mappings can be toggled using "monitor toggleextmemaccess" instead (issue reported by @Nanquitas). Closes#943.
Additionally, support for the buggy IDA 6.8's buggy gdb client was removed. Please use IDA 7.0 or higher.
P.S: IDA 7.0 added client support for no-ack mode (Debugging options > Set specific options), which should be enabled to make debugging twice as smooth. "Use stepping support" should be unchecked.
Instead of being called top_XXXX.bmp where XXXX means nothing,
screenshots are now called YYYY-MM-DD_HH-MM-SS.mmm_top.bmp (and same
idea for bot and top_right).
First obvious consequence, this is easier to manage in a file browser.
Now you have screenshots grouped by time, and you also have the 2 or 3
related screenshots (top, bot, top_right) close one to another.
Another consequence is that there is no need to go through existing
screenshots to find an unused number. Now osGetTime takes care of
everything. And going through files was a very time consuming process.
Now screenshots only need 4 (or 6 if 3D) seconds to be saved (instead of
40 seconds if you had 200 screenshots in your directory already).
Conflicts may happen when people change the date and time on their 3DS,
but that's why I even included milliseconds in the timestamp. People
don't set date and time everyday and they'd be unlucky to take a
screenshot at the exact time, milliseconds included, as another
screenshot right after setting the time one hour earlier...
Source for converting seconds since 1970 (or 1900...) to date and time:
https://stackoverflow.com/questions/21593692/
Virtually full support for 5.x except that svcUnmapProcessMemoryEx will be forwarded to svcUnmapProcessMemory (both are equivalent for up to 64MB chunks)
