Investigative Reports on Certain micro SD card brands failing

[Dartz150]

Users have reported frequent system freezes and loading failures while performing certain actions while using the DSpico and its components. While initial suspicions focused on the DSpico hardware, the root cause remains unconfirmed, because the issue can be also reproduced in other flashcards while trying to launch anything with pico-launcher. We are now investigating a possible conflict within any of the DSPico components or the interaction between the hardware's I/O and specific MicroSD controllers.

This generic issue will centralize this research for now and will be moved accordingly after we find the project module that has the problem.

Before you submit a report:

• Check if your SD card isn't a fake using this guide.
• Perform an integrity test using HW2Testw.

Please consider using this template while reporting your problematic SD card:

- **Exact Brand & Model:** [e.g., SanDisk Extreme / Samsung EVO]
- **Capacity:** [e.g., 32GB / 256GB]
- **Technical Specifications:**
  - **Speed Class:** [e.g., Class 10 / U3 / V30]
  - **App Performance Class:** [e.g., A1 / A2]
  - **SD Formatter Tool:** [e.g., SD Formatter/GUI Format]
  - **Allocation Unit Size (Cluster):** [e.g., 32kb / 64kb]
- **Environment Details:**
  - **DSPico Hardware Revision:** [e.g., v1.3]
  - **DSPico Firmware Version:** [e.g., v1.0.0]
  - **Pico Launcher Version:** [e.g., v1.2.1]
  - **Pico Loader Version:** [e.g., v1.2.1]
  - **Pico Bootloader Version:** [e.g., v1.0.0]
  - **Pico DLDI Version:** [e.g., v1.0.1]
  - **Flashcard Platform (Only if you're not using the DSpico):** [e.g., Ace3DS]
  - **Trigger Action:** [e.g., Navigating menu / Launching .nds homebrew or retail / During gameplay]
- **Error Description:** [e.g., Black screen / Total freeze / "Card not found" error]
- **Integrity Check (H2testw/F3):** [Pass / Fail / Not Tested]

It would also help if you post a picture of your SD card. Any report that fails to provide the requested details will be ignored.

[Dartz150]

• Exact Brand & Model: "Samsung PRO"
• Capacity: 32GB
• Technical Specifications:
  • Speed Class: Class 10
  • App Performance Class: N/A
  • SD Formatter Tool: SD Card Formatter
  • Allocation Unit Size (Cluster): 32kb
• Environment Details:
  • DSPico Hardware Revision: v1.2
  • DSPico Firmware Version: v1.0.0
  • Pico Launcher Version: v1.2.1
  • Pico Loader Version: v1.3.1
  • Pico Bootloader Version: v1.0.0
  • Pico DLDI Version: v1.0.1
  • Flashcard Platform (Only if you're not using the DSpico): Ace3DS, TTDS, DSpico
  • Trigger Action: Trying to load anything while in pico-launcher.
• Error Description: The system freezes, the launcher plays the BG music in a weird loop, you're forced to turn off the console.
• Integrity Check (H2testw/F3): Pass.

[ApacheThunder]

I am pretty sure this is something in the DSPico Firmware or it's DLDI. I have a card that has this issue where Pico Launcher hangs bootings things sometimes. BUT this only occurs on DSPico! I swapped out this card for a 4GB SanDisk card that used to be what my Acekard 2i was using and it works fine on that now. (so not all SanDisk cards have this problem. Maybe the older ones are still fine. The working Sandisk card I have is a 4GB one. Same capacity and formatted to same cluster size as the non working one too.

Anyways the non working card was swapped over to my Acekard 2i where I also setup Pico Loader/Launcher on. But here's the thing. It works fine on that flashcart despite nothing about the microSD card being changed besides the Pico Loader platform using AK2i DLDI and such!

So pretty sure it's something to do with DS Pico and it's firmware. Not so much what ever version of Pico Loader/Launcher you happen to be using.

Anyways I don't recall what version of Pico Loader I had on the Acekard SD card when I retested this. The non working card is an offbrand of unknown origin. I don't remember where I go this one from. I think it came with one of the other flashcarts I got over the years. I'm pretty sure it's not a fake. 4GB is too small a capacity for some random Chinease vender to "fake" anyways. :P

I've never had issues with this card either. Only with DS Pico did it have any problems.

Anyways some photos of the non working card:

And a photo of the working one:

I'm not sure what class card that one is. Class I maybe?

Anyways some things to note about the non working one. It has noticeably faster read/writes then my working Sandisk card. So the class 10 rating on it might be real at least. :P

Maybe this issue crops up specifically in class 10 cards? Does anyone else have class 10 cards that work fine?

Oh and both cards are formatted fat32 with 32KB cluster size. Don't remember if I formatted them both with SD formatter or not.

[chreasyyy]

• Exact Brand & Model: Verbatim Premium (Model 44010)
• Capacity: 16GB
• Technical Specifications:
  • Speed Class: Class 10
  • App Performance Class: idk
  • SD Formatter Tool: I tried both diskutil & SD Card Formatter
  • Allocation Unit Size (Cluster): 32kb
• Environment Details:
  • DSPico Hardware Revision: v1.3
  • DSPico Firmware Version: v1.0.0
  • Pico Launcher Version: v1.0.0
  • Pico Loader Version: v1.3.1
  • Pico Bootloader Version: v1.0.0
  • Pico DLDI Version: v1.0.1
  • Flashcard Platform (Only if you're not using the DSpico): DSPico
  • Trigger Action: Launching any .nds file
• Error Description: Total freeze (2 out of 3 times)
• Integrity Check (H2testw/F3): Pass

[davidgfnet]

I've been investigating this with Gericom's help, and I think I found the issue. There's some bug in the SD card TX FSM, which causes it to do weird stuff in some cases. My current fix, which needs significant testing (in both "good" and "bad" SD cards), looks like:

diff --git a/src/sd/rp2040_sdio.cpp b/src/sd/rp2040_sdio.cpp
index 1b75392..b9d4813 100644
--- a/src/sd/rp2040_sdio.cpp
+++ b/src/sd/rp2040_sdio.cpp
@@ -662,6 +662,7 @@ static void rp2040_sdio_irq()
                 channel_config_set_dreq(&dmacfg, pio_get_dreq(SDIO_PIO, SDIO_DATA_SM, false));
                 dma_channel_configure(SDIO_DMA_CHB, &dmacfg,
                     &g_sdio.card_response, &SDIO_PIO->rxf[SDIO_DATA_SM], 1, true);
+                return;
             }
         }
     }

[davidgfnet]

For completeness my understanding of the issue is:

• Most of the time "rp2040_sdio_irq" is called and things flow via "// Card is already idle", so the write proceeds to the next block (or is completed).
• Whenever things go via the other branch "// Use DMA to wait for the response", it schedules a DMA and, in some cases, falls into the "g_sdio.transfer_state == SDIO_TX_WAIT_IDLE" state. That depends on the gating "dma_channel_is_busy" below. There's probably some race conditions there.
• I suspect that should never happen and, after scheduling the DMA, we should finish the IRQ handler and wait for the next IRQ call. Not doing so results in two IRQs being generated and g_sdio.blocks_done overflowing over g_sdio.total_blocks, which hangs forever (we always validate "if (g_sdio.transfer_state == SDIO_TX_WAIT_IDLE && g_sdio.blocks_done == g_sdio.total_blocks)", which will never be true after the overflow).

I saw some evidence of the aforementioned "done" overflow, also about the double "IRQ". Adding this return seems to avoid the overflow and the completion is handled in the second IRQ event.

[ApacheThunder]

I implemented your change and so far haven't seen it hang again on the SD card that wasn't working correctly. I made it my main SD card for my DSPico now so if it does happen again I'll let you know but it looks like it might have solved it for me.

[ORelio]

Was eager to try my brand new DSpico, but it took me five(!!) attempts to find a working 64GB SD card 🤯

• Exact Brand, Model, Capacity:
  • SanDisk Ultra 64GB [FAIL]
  • thinkplus 64GB [FAIL]
  • SanDisk Extreme Plus 32GB [PASS]
  • Kingston Canvas Select Plus 64GB [FAIL]
  • Kodak 64 GB [PASS]
• Technical Specifications:
  • Speed Class: U1 or U3
  • App Performance Class: A1
  • SD Formatter Tool: Gparted, MBR, one FAT32 partition, entire SD card
  • Allocation Unit Size (Cluster): 32K (default setting)
• Environment Details:
  • Firmware Build Environment: Ubuntu with Docker as per Video and Dockerfile
  • DSPico Hardware Revision: v1.3 (ChisFamily, TRN BK NO TF)
  • DSPico Firmware Version: Dockerfile cloned develop at commit 117b25a on 02/02/2026
  • Pico Launcher Version: v1.1.0, manual download and copied to SD cards
  • Pico Loader Version: v1.5.0, manual download and copied to SD cards
  • Pico Bootloader Version: Dockerfile cloned develop, still at tag v1.0.0 on 02/02/2026
  • Pico DLDI Version: Dockerfile cloned develop, still at tag v1.0.1 on 02/02/2026
  • Flashcard Platform (Only if you're not using the DSpico): N/A
  • Trigger Action: Insert DSpico in a DS Lite or DSi and attempt to boot it
• Error Description:
  • SanDisk, thinkplus: White screen past launch (DS Lite) or wrfuxxed exploit (DSi)
  • Kingston: SD seems not recongnized by DSpico as the cart does not appear in DSi menu
• Integrity Check (H2testw/F3): PASS for all five cards

All cards formatted with gparted, FAT32, default settings, then tested successfully with F3, and put in use with other devices. I only had the SanDisk Ultra 64GB laying around but it did not work, had to swap SDs with other devices to find one that worked with DSpico, so all the others were already in-use elsewhere with no issues.

[soulpower11]

Tested David's changes with my card (Samsung Evo Plus 32GB) that wasn't working.

Haven't encountered any freezing/hang while loading up my games since then. Looks like his fix solved the problem

[ORelio]

Applying @davidgfnet's proposed change over commit 117b25a then recompiling firmware:

... Way better indeed. No more white screens 🎉

The only SD card that does not pass on my side seems to be not recognised at all, but not a big deal.
I could switch to the SanDisk Ultra I initially planned to use.
Will let you know if I face further issues.

Thanks for your work on this cool project!

[chreasyyy]

I just tested davids fix for an hour and there were no crashes at all, my sd card works perfectly fine now!

[puncle]

• Exact Brand & Model: Lexar E-Series
• Capacity: 32gb
• Technical Specifications:
  • Speed Class: Class 10
  • App Performance Class: A1
  • SD Formatter Tool: SD Card Formatter
  • Allocation Unit Size (Cluster): 32kb
• Environment Details:
  • DSPico Hardware Revision: v1.3 (ChisFamily)
  • DSPico Firmware Version: v1.0.0
  • Pico Launcher Version: v1.1.0
  • Pico Loader Version: v1.5.0
  • Pico Bootloader Version: v1.0.0
  • Pico DLDI Version: v1.0.1
  • Flashcard Platform (Only if you're not using the DSpico): N/A
  • Trigger Action: Launching the card from DS Lite menu
• Error Description: White screen on launch
• Integrity Check (H2testw/F3): Pass

[BlazeVertex]

• Exact Brand & Model: [SanDisk Extreme]
• Capacity: [64GB]
• Technical Specifications:
  • Speed Class: [U3 / V30]
  • App Performance Class: [A2]
  • SD Formatter Tool: [GUI Format]
  • Allocation Unit Size (Cluster): [32kb]
• Environment Details:
  • DSPico Hardware Revision: [ v1.3 (ChisFamily)]
  • DSPico Firmware Version: [1.0.0]
  • Pico Launcher Version: [v1.1.0]
  • Pico Loader Version: [1.5.0]
  • Pico Bootloader Version: [ v1.0.0]
  • Pico DLDI Version: [v1.0.1]
  • Flashcard Platform (Only if you're not using the DSpico): [e.g., Ace3DS]
  • Trigger Action: [When just opening to go into the launcher itself]
• Error Description: [ first white than black and nothing after that, also with the same sd card it works on my unmoded 2DS but does not on my Modded 3DS]
• Integrity Check (H2testw/F3): [Not Tested (brand new sd card and will prob don't show any signs of problems]
• wrfuxed was used to test it and also tried this patch

I can provide a video if needed.

[chr0m]

Nice, I added the fix from davidgfnet and now cards that were previously freezing with white screens are working

[davidgfnet]

The PR was merged in dspico-firmware, should we close this perhaps? Most folks already confirmed this fixes the issue for them. Any non-working cards remaining should be thoroughly tested, it could be they are broken?
If someone wants to send me a bad cart I have the necessary hardware tooling to debug it, but no promises I can find the issue (if there's an issue!)

[BlazeVertex]

The PR was merged in dspico-firmware, should we close this perhaps? Most folks already confirmed this fixes the issue for them. Any non-working cards remaining should be thoroughly tested, it could be they are broken?
If someone wants to send me a bad cart I have the necessary hardware tooling to debug it, but no promises I can find the issue (if there's an issue!)

#4 (comment)

Can you maybe answer this quick question why does the dspico on my unmodded o2DS work with the old and new firmware with that SD card but not on my modded old 3DS xl