Register table
Old3DS | Name | Address | Width | Used by |
---|---|---|---|---|
Yes | PDN_SLEEP_CNT | 0x10141000 | 2 | Kernel11 |
Yes | PDN_WAKE_ENABLE | 0x10141008 | 4 | PTM Services, PDN Services |
Yes | PDN_WAKE_REASON | 0x1014100C | 4 | PTM Services, TwlBg, PDN Services |
Yes | LGY_MODE | 0x10141100 | 2 | TwlProcess9, TwlBg |
Yes | LGY_SLEEP | 0x10141104 | 2 | TwlBg |
Yes | LGY_IRQ_ENABLE | 0x10141108 | 2 | TwlBg |
Yes | LGY_PADCNT | 0x1014110A | 2 | TwlBg |
Yes | LGY_HIDEMU_MASK | 0x10141110 | 2 | TwlBg |
Yes | LGY_HIDEMU_PAD | 0x10141112 | 2 | TwlBg |
Yes | LGY_GPIOEMU_MASK | 0x10141114 | 2 | Codec Services, TwlBg |
Yes | LGY_GPIOEMU_DATA | 0x10141116 | 2 | Codec Services, TwlBg |
Yes | LGY_CARDDETECTEMU_MASK | 0x10141118 | 1 | TwlBg |
Yes | LGY_CARDDETECTEMU_DATA | 0x10141119 | 1 | TwlBg |
Yes | LGY_? | 0x10141120 | 1 | TwlBg |
Yes | PDN_GPU_CNT | 0x10141200 | 4 | Boot11, Kernel11, PDN Services, TwlBg |
Yes | PDN_VRAM_CNT | 0x10141204 | 1 | Boot11, Kernel11, TwlBg |
Yes | PDN_LCD_CNT | 0x10141208 | 1 | Boot11 |
Yes | PDN_FCRAM_CNT | 0x10141210 | 2 | Kernel11, TwlBg |
Yes | PDN_I2S_CNT | 0x10141220 | 1 | Boot11, TwlBg, PDN Services |
Yes | PDN_CAMERA_CNT | 0x10141224 | 1 | PDN Services |
Yes | PDN_DSP_CNT | 0x10141230 | 1 | Process9, PDN Services |
No | PDN_MVD_CNT | 0x10141240 | 1 | |
No | PDN_LGR_SOCMODE | 0x10141300 | 2 | NewKernel11 |
No | PDN_LGR_CNT | 0x10141304 | 2 | NewKernel11 |
No | PDN_LGR_CPU_CNT<0-3> | 0x10141310 | 1*4 | NewKernel11 |
Sleep registers
PDN_SLEEP_CNT
Bits | Description |
---|---|
0 | 1 = Enter sleep mode |
2-14 | Unused. |
15 | 1 = VRAM is powered down |
Kernel11 powers down VRAM (it's unclear whether bit15 is power-down or self-refresh mode) by first disabling the 8 banks using GX register 0x10400030, then by disabling the GPU clock using PDN_GPU_CNT bit 16 and finally writes to and polls this register.
PDN_WAKE_ENABLE
Bits | Description |
---|---|
1 | HID_PADCNT |
3 | Shell opened |
4 | Headphones not plugged in |
8 | WiFi (?) |
19 | Shell GPIO (?) |
26 | MCU interrupt |
30 | Touch screen pressed |
31 | CTR gamecard inserted/removed |
List in progress.
This is a OR list of wake triggers that will wake up the console from sleep and raise IRQ 0x58.
PDN_WAKE_REASON
Same layout as PDN_WAKE_ENABLE.
This is a OR list of the wake triggers that actually woke up the console.
For each bit, write 1 to acknowledge, and 0 to clear (?).
Legacy registers
LGY_MODE
Bits | Description |
---|---|
0-1 | Read only legacy mode set on reg 0x10018000. |
2-14 | Unused. |
15 | 1 = enable legacy mode. |
To boot into DSi or GBA mode first set register 0x10018000 to the desired mode and setup LgyFb. Then disable FCRAM by clearing bit 0 in reg 0x10201000, writing 0 to PDN_FCRAM_CNT followed by 1 and waiting for bit 2 to clear.
The very last 3DS-mode register poke the TWL_FIRM Process9 does before it gets switched into TWL-mode, is writing 0x8000 to this register. Before writing this register, TWL Process9 waits for ARM7 to change the value of this register. The Process9 code for this runs from ITCM, since switching into TWL-mode includes remapping all ARM9 physical memory.
LGY_SLEEP
Bits | Description |
---|---|
0 | Write 1 to wakeup GBA mode. |
1 | Sleep state/ack. 1 when GBA mode entered sleep. Write 1 to ack. |
2 | ? |
3-14 | Unused. |
15 | 1 = IRQ enable (IRQ 0x59) |
When a GBA game enters sleep mode and bit 15 is 1, IRQ 0x59 fires and bit 1 is set. Bit 1 must be acknowledged/written together with bit 0 otherwise GBA mode wakes up from sleep early sometimes.
LGY_IRQ_ENABLE
Arm11 interrupt enable bits for legacy interrupts, same bit layout as the GPIOEMU regs below.
LGY_PADCNT
Also named "KEYCNT" on certain other DS(i)/GBA documentations. The value of this register is copied to HID_PADCNT when GBA mode enters sleep.
LGY_HIDEMU_MASK
Set bits will use the corresponding values from LGY_HIDEMU_PAD instead of allowing the hardware to read it from HID_PAD.
This is set to 0x1FFF (all buttons and the debug key) and LGY_HIDEMU_PAD is set to 0 when the "Close this software and return to HOME Menu?" dialog is shown to prevent the button presses from propagating to the DS/GBA CPU.
LGY_HIDEMU_PAD
Works the same way as HID_PAD, but the values set here are only replaced in the HID_PAD seen by the DS/GBA CPU when the corresponding bits in LGY_HIDEMU_MASK are set.
LGY_GPIOEMU_MASK
Set bits will read bits from LGY_GPIOEMU_DATA (override).
This is used to trigger things like the TWL MCU interrupt in TWL mode.
LGY_GPIOEMU_DATA
See above
LGY_CARDDETECTEMU_MASK
Set bits will read bits from LGY_CARDDETECTEMU_DATA (override).
Bit0 signals cartridge removal.
LGY_CARDDETECTEMU_DATA
See above
Clock and reset registers
PDN_GPU_CNT
Bits | Description |
---|---|
0 | GPU main block + VRAM + LCD reset. 0 = reset. |
1 | PSC block reset? 0 = reset. |
2 | Geoshader block reset? 0 = reset. |
3 | Rasterization block reset? 0 = reset. |
4 | PPF block reset. 0 = reset. |
5 | PDC block reset? 0 = reset. |
6 | PDC related reset. 0 = reset. |
7-15 | Unused. |
16 | Clock enable for all blocks, VRAM and LCD. 1 = enable. |
PDN uses a 12 ARM11 cycle delay to deassert reset.
PDN_VRAM_CNT
Bits | Description |
---|---|
0 | Clock. 1 = enable, 0 = disable |
This register seems to be unimplemented in released models: while it is used in tandem with PDN_GPU_CNT.bit16 in Boot11 screeninit code, Kernel11 only uses PDN_GPU_CNT.bit16 to power-off VRAM before going to sleep.
PDN_LCD_CNT
Bits | Description |
---|---|
0 | Clock. 1 = enable, 0 = disable |
This register seems to be unimplemented in released models, only to be used in Boot11, as PDN_GPU_CNT.bit16 also drives the LCD clock.
PDN_FCRAM_CNT
Bits | Description |
---|---|
0 | Reset. 0 = reset. |
1 | Clock. 1 = enable, 0 = disable |
2 | Acknowledge clock request. Gets set or unset when toggling bit 1. |
Twl-/AgbBg use this to disable FCRAM for the GBA rom in GBA mode or DSi main RAM in DSi mode. AgbBg clears bit 0 in reg 0x10201000 before touching this reg.
Kernel11 uses it to put the FCRAM in self-refresh mode (clock disable) before going to sleep.
PDN_I2S_CNT
Bits | Description |
---|---|
0 | I2S1 Clock (maybe?) 1 = enable, 0 = disable |
1 | I2S2 Clock. 1 = enable, 0 = disable |
I2S1 clock enable bit seems to be unimplemented. Maybe it's because DSP clock enable drives it?
PDN_CAMERA_CNT
Bits | Description |
---|---|
0 | Clock. 1 = enable, 0 = disable |
PDN_DSP_CNT
Bits | Description |
---|---|
0 | Reset. 0 = reset. |
1 | Clock. 1 = enable, 0 = disable |
PDN services holds reset for 0x30 Arm11 cycles.
PDN_MVD_CNT
Bits | Description |
---|---|
0 | Reset. 0 = reset |
This doesn't seem to be used by anything, but does have a clear effect on the hardware.
The reset value for this register is 1 (out-of-reset at boot).
N3DS SoC (LGR) registers
PDN_LGR_SOCMODE
This is used for configuring the New3DS ARM11 CPU clock-rate. This register is New3DS-only: reading from here on Old3DS always returns all-zeros even when one tried writing data here prior to the read.
Bits | Description |
---|---|
0-2 | SoC mode.
Possible values: 0=CTR+256MHz 1=LGR2+256MHz, 5=LGR2+804MHz 2=LGR+256MHz, 3=LGR+536MHz N3DS modes (LGR1/LGR2) enable the New 3DS FCRAM extension and are needed to access N3DS-only devices.
|
15 | Busy |
On firmlaunch, the kernel sets the mode to O3DS.
svcKernelSetState type10, only implemented on New3DS, uses this register. That code writes the following values to this register, depending on the input Param0 bit0 state, and the state of CFG11_SOCINFO:
Register value | Higher-clockrate bit set in svcKernelSetState Param0 | CFG11_SOCINFO bit2 set | MPCore timer/watchdog prescaler value, prior to subtracting it by 0x1 when writing it into hw/state | Clock-rate multiplier | Description |
---|---|---|---|---|---|
0x01 | No | Yes | 0x01 | 1x | 268MHz |
0x02 | No | No | 0x01 | 1x | 268MHz |
0x05 | Yes | Yes | 0x03 | 3x | 804MHz |
0x03 | Yes | No | 0x02 | 2x | 536MHz (tested on New3DS) |
Note that the above CFG11_SOCINFO bit is 1 on New3DS, and 0 on Old3DS. Since this SVC is only available with the New3DS ARM11-kernel, the only additional available clock-rate is 804MHz when running on New3DS(with official kernel code).
The following register value(s) were tested on New3DS by patching the kernel:
- 0x00: Entire system hangs.
- 0x02: Entire system hangs.
- 0x03: ARM11 runs at 536MHz.
- 0x04: Entire system hangs.
- 0x06: Entire system hangs.
- 0x07: Same result as 0x05.
- 0x08: Entire system hangs.
- 0x09: Entire system hangs.
- 0x0A: Entire system hangs.
- 0x0B: Same result as 0x03.
- 0x0C: Entire system hangs.
- 0x0D: Same result as 0x05.
- 0x0E: Entire system hangs.
- 0x0F: Same result as 0x05.
- 0x1F, 0x2F, 0x4F, 0x8F, 0xFF: Same result as 0x05.
PDN_LGR_CNT
Bits | Description |
---|---|
0 | Enables the N3DS extramem block |
8 | Enables the L2C block (LGR2 only) |
Kernel11 sets this to 0x101 when bit 2 in CFG11_SOCINFO (LGR2 supported) is set otherwise 1.
PDN_LGR_CPU_CNT<0-3>
Bits | Description |
---|---|
0 | nRESET, 0 = reset. Also enable the bootrom instruction overlay. |
1 | Enable bootrom data overlay |
4 | Reset operation in progress |
5 | Always 1? |
Only usable for core2 and core3.
The normal Arm11 bootrom checks cpuid and hangs if cpuid >= 2. This is a problem when booting the 2 additional New3DS Arm11 MPCores. NewKernel11 solves this by using a hardware feature to overlay the bootrom with a configurable branch to a kernel function. This overlay feature was added with the New3DS.
Bit1 in register above enables a bootrom data-override for physical addresses 0xFFFF0000-0xFFFF1000 and 0x10000-0x11000. All _data reads_ made to those regions now read the 32-bit value provided in CFG11_BOOTROM_OVERLAY_VAL.
Bit0 sets the core out of reset and enables a bootrom instruction-overlay which means that _instruction reads_ made to the bootrom region are overridden. We have not been able to dump what instructions are actually placed at bootrom by this switch (because reading the area only yields data-reads). Jumping randomly into the 0xFFFF0000-0xFFFF1000 region works fine and jumps to the value provided by the data overlay CFG11_BOOTROM_OVERLAY_VAL. Thus we may predict that the entire bootrom region is filled by: ldr pc, [pc]
Or equivalent. However, jumping to some high addresses such as 0xFFFF0FF0+ will crash the core. This may be explained by prefetching in the Arm pipeline, and might help us identify what instructions are placed by the instruction-overlay.