Puppy Linux Discussion Forum

Solution found, a boot parameter can fix this

`"sdhci.debug_quirks=0x10000"`

Look [here](https://forum.puppylinux.com/viewtopic.php?p=21958#p21958) for some considerations

---

Hello everybody!
I'm encountering an issue with the micro sd-card reader on my tablet/mini-notebook.
I've tried running *fossapup64-9.5* and *Fatdog64 811 Final* on my `Acer Switch 10 SW5-012-13TT`. It's one of those cheap ablets with windows 8.1 and atom 64bit processor, but mounted with a 32bit OS and uefi... I can't complain it's quite sturdy and performant for browsing and reading and also viewing couple videos.

As of now both puppies run nicely with minor annoyances, my main issue is the internal micro sd-card reader: no matter what card, what type of formatting (fat, ntfs, ext#, you name it), the system always mount the card partition as "read-only". I can read and write any cards in windows.

I mostly care for the sd card reader because I wanted to use an external memory for puppy savefiles and savedirs. Unfortunately this tablet has a very small 20gb SSD with windows installed filling it almost to max, and I'd like to get more confident with linux before formatting the entire disk.

I've tried to fiddle around a bit with scarce results, but my linux proficiency is close to none.
Searching around I've found posts suggesting various info commands to try to debug this, I think the most relevant are the following

```
# fdisk -list
Disk /dev/mmcblk2: 59.5 GiB, 63879249920 bytes, 124764160 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: dos
Disk identifier: 0x00000000

Device Boot Start End Sectors Size Id Type
/dev/mmcblk2p1 32768 124764159 124731392 59.5G 7 HPFS/NTFS/exFAT

Disk /dev/mmcblk1: 29.1 GiB, 31272730624 bytes, 61079552 sectors
Units: sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disklabel type: gpt
Disk identifier: XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX

Device Start End Sectors Size Type
/dev/mmcblk1p1 2048 206847 204800 100M EFI System
/dev/mmcblk1p2 206848 468991 262144 128M Microsoft reserved
/dev/mmcblk1p3 468992 44300287 43831296 20.9G Microsoft basic data
/dev/mmcblk1p4 44300288 61077503 16777216 8G Windows recovery environment
```

```
# mount
/dev/mmcblk2p1 on /mnt/mmcblk2p1 type fuseblk (ro,relatime,user_id=0,group_id=0,default_permissions,allow_other,blksize=4096)
/dev/mmcblk1p3 on /mnt/mmcblk1p3 type fuseblk (rw,relatime,user_id=0,group_id=0,default_permissions,allow_other,blksize=4096)
```

I've read an ubuntu thread where it has been suggested this could be a firmware/driver issue maybe, but I can't find right now.

If you need more info let me know.

In any case, let me say thank you to the entire puppy community for your incredible effort and commitment to help newcomers and keeping this incredible resource of knowledge and amazing OS!

I have never seen this problem.

Even if it works OK in Windows.
I would check the UEFI bios setup for any settings that could affect the drive.
I suggest you have one of these settings in the UEFI setup. (depends on UEFI what it offers)
secure boot disabled or legacy boot enabled or CSM enabled

Are you using Windows program to partition and format?
If yes.
Try using Gparted program, in Puppy, to partition and format.
I suggest to completely start fresh.
Make partition table (msdos)
Partition and format.

Shut down Windows normally, as a complete shutdown.
Do not shut down in hibernate.
Hibernate sets up drives different from normal for boot up.

Since all your disk devices appear to be MicroSD cards, I'm going to guess that /dev/mmcblk2 is your external 64GB MicroSD card formatted in fat32 possibly?

I apologize in advance, but it has to be asked; Have you made sure the "read only" lock hasn't inadvertently been set on the MicroSD?

Partition mmcblk2p1 has an odd starting position.

I would follow bigpup's advice to run Gparted, create a new partition table (under the Device tab) and make some new partitions.

Do you have an external usb to SD card reader that you could try? It would be interesting to know if you got the same messages. If your problem is caused by a firmware issue with the internal reader then using an external reader may work ok.

Thank you all for your time and answers!

I'll answer to @Jafadmin and @greengeek first since it should be quick.

@Jafadmin: sorry I should have specified with more clarity that the card reader slot is for micro sd-cards only, I can't fit that big adapter.
I should have also mentioned more clearly what the two disks are.
You are correct in your assumption that disks appear as micro sd-cards but only mmcblk2 is really the micro sd-card I'm having issue with. mmcblk1 is the internal SSD memory of the tablet with windows, so not really an sd-card if I understand correctly what mmcblk devices are.

@greeengeek I've tried with one of those usb dongles with sd-card slots and you are correct it works as expected, read and write operations are available. Unfortunately on this tablet I have only 1 usb-mini slot available on the tablet itself, and 1 standard usb port on the detachable keyboard but I use it already to plug the usb stick with loaded puppy, so it's doable but pretty clunky as I should use a mini-usb dongle to use the sd-card.

@bigpup and @rcrsn51: UEFI options and settings are scant unfortunately, the secure boot is already disabled, and I've tested also without fastboot/hybernate options, full shutdown without success. I think windows is not tampering with the sd-card because as I reported above I can read and write the windows partitions.
I've tried your suggestions in formatting a 8gb sd-card directly in gparted but it failed for the same reason: read only device. I've tried also to format the card with a usb dongle, and was able to do so as ext3, but again once inserted in the sd-card slot was mounted as read-only device.
Tried again to format while in the sd-card slot from ext3 to ext4 but no dice, the last gparted-details.html content is reported below, but all previous are same.

Are there any options left?

Thank you all again!

GParted 1.0.0

configuration --enable-libparted-dmraid --enable-online-resize

libparted 3.3

========================================

Device: /dev/mmcblk1
Model: MMC HBG4e
Serial: none
Sector size: 512
Total sectors: 61079552  
Heads: 255
Sectors/track: 2
Cylinders: 119763  
Partition table: gpt
 

  Partition        Type      Start      End        Flags          Partition Name                 File System   Label    Mount Point
  ---------------- --------- ---------- ---------- -------------- ------------------------------ ------------- -------- -------------
  /dev/mmcblk1p1   Primary   2048       206847     boot, esp      EFI system partition           fat32         ESP      
  /dev/mmcblk1p2   Primary   206848     468991     msftres        Microsoft reserved partition   unknown                
  /dev/mmcblk1p3   Primary   468992     44300287   msftdata       Basic data partition           ntfs          Acer     
  /dev/mmcblk1p4   Primary   44300288   61077503   hidden, diag   Basic data partition           ntfs          Images   

========================================

Device: /dev/mmcblk2
Model: SD SA08G
Serial: none
Sector size: 512
Total sectors: 15130624  
Heads: 255
Sectors/track: 2
Cylinders: 29667  
Partition table: msdos
 

  Partition        Type      Start   End        Flags   Partition Name   File System   Label   Mount Point
  ---------------- --------- ------- ---------- ------- ---------------- ------------- ------- -------------
  /dev/mmcblk2p1   Primary   8192    15130623                            ext3                  

========================================

**Format /dev/mmcblk2p1 as ext4**  00:00:06    ( ERROR )     

calibrate /dev/mmcblk2p1  00:00:00    ( SUCCESS )     

*path: /dev/mmcblk2p1 (partition)\
start: 8192\
end: 15130623\
size: 15122432 (7.21 GiB)*

libparted messages    ( INFO )     

*Unable to open /dev/mmcblk2 read-write (Read-only file system).
/dev/mmcblk2 has been opened read-only.*

clear old file system signatures in /dev/mmcblk2p1  00:00:06    ( ERROR )     

write 512.00 KiB of zeros at byte offset 0  00:00:06    ( ERROR )     

libparted messages    ( ERROR )     

*Unable to open /dev/mmcblk2 read-write (Read-only file system).
/dev/mmcblk2 has been opened read-only.*

*Unable to open /dev/mmcblk2 read-write (Read-only file system).
/dev/mmcblk2 has been opened read-only.*

*Can't write to /dev/mmcblk2, because it is opened read-only.*
*Can't write to /dev/mmcblk2, because it is opened read-only.*

It looks like this card slot has some kind of hardware incompatibility with Linux.

Run the command: dmesg | grep -i mmc

For what its worth in case helpful I mounted an SD micro in the usb slot of my desktop and ran GParted on it .The screenshot be;ow shows how mine comes up . I'm using one of those small micro to USB adapters so maybe you could get one and try it in USB.
GParted shows it with a strange lock on it sayin LBA flag I don't understand but I do not seem to be able to write to it . I'm using xenialpup64-7.5
Its mounted as read only too.

root# dmesg | grep -i mmc
PCI: MMCONFIG for domain 0000 [bus 00-ff] at [mem 0xe0000000-0xefffffff] (base 0xe0000000)
PCI: MMCONFIG at [mem 0xe0000000-0xefffffff] reserved in E820
wbsd: Winbond W83L51xD SD/MMC card interface driver    <------------------------------------------------------------------------*
sr 1:0:0:0: [sr0] scsi3-mmc drive: 24x/24x writer cd/rw xa/form2 cdda tray
root# 

By unmounting it I then opened it in GParted and formatted it to Fat32 .
Then I mounted it again and was able to successfully write to it .

My card came out of an IP camera to start so maybe that is writing it in an unusual way.

My card came out of an IP camera to start so maybe that is writing it in an unusual way.

The clue to solving the problem.
Yes.
Cameras do formatting in not normal ways.
It is the way the specific camera wants it done.
It may show as fat32, but the format has block size, and other settings the way the camera wants them.
The specific camera can write to it, but not other devices.
Plus who knows if a special security code got put on the drive, that only the specific camera will unlock the drive for writing.

Usually doing it on a computer.
Making a new partition table, partitioning, and formatting gets it back to normal setup.
Key word usually.
Sometime it may require more rigorous terminal commands to clear out what the camera put on the SD card.

As @rcrsn51 mentioned here below is the dmesg output without the sd-card inserted on FatDog64

# dmesg | grep -i mmc
[    0.112749] PCI: MMCONFIG for domain 0000 [bus 00-3f] at [mem 0xe0000000-0xe3ffffff] (base 0xe0000000)
[    0.112769] PCI: not using MMCONFIG
[    1.828095] PCI: MMCONFIG for domain 0000 [bus 00-3f] at [mem 0xe0000000-0xe3ffffff] (base 0xe0000000)
[    1.834223] [Firmware Info]: PCI: MMCONFIG at [mem 0xe0000000-0xe3ffffff] not reserved in ACPI motherboard resources
[    1.834229] PCI: not using MMCONFIG
[    1.865512] acpi PNP0A08:00: fail to add MMCONFIG information, can't access extended PCI configuration space under this bridge.
[    3.479098] wbsd: Winbond W83L51xD SD/MMC card interface driver
[    8.019731] mmc0: SDHCI controller on ACPI [INT33BB:00] using ADMA
[    8.034602] mmc1: SDHCI controller on ACPI [80860F14:00] using ADMA
[    8.047840] mmc2: SDHCI controller on ACPI [80860F14:01] using ADMA
[    8.067112] mmc0: new high speed SDIO card at address 0001
[    8.196432] mmc1: new HS200 MMC card at address 0001
[    8.198211] mmcblk1: mmc1:0001 HBG4e 29.1 GiB 
[    8.198471] mmcblk1boot0: mmc1:0001 HBG4e partition 1 4.00 MiB
[    8.198807] mmcblk1boot1: mmc1:0001 HBG4e partition 2 4.00 MiB
[    8.198975] mmcblk1rpmb: mmc1:0001 HBG4e partition 3 4.00 MiB, chardev (246:0)
[    8.201781]  mmcblk1: p1 p2 p3 p4

and here the result after inserting the sd-card in the slot

# dmesg | grep -i mmc
[  103.507324] mmc2: new high speed SDXC card at address 0001
[  103.508327] mmcblk2: mmc2:0001 00000 59.5 GiB (ro)
[  103.512452]  mmcblk2: p1

Interestingly I see the same line as @April in the dmesg output.

wbsd: Winbond W83L51xD SD/MMC card interface driver

But I guess the interesting parts are the first lines where it complains about mmconfig but unfortunately I can't really tell what it means. With some really quick googling I found this thread but I don't know if it applies at all to my situation.

[Resolved] No extended PCIe config space access - CentOS

Considering @bigpup point on cameras formatting, I'm using some standard sd-cards bought online, I'm not reusing cards coming from other devices. If it matters I've tested with a 64gb samsung card and some old cheap 8gb kingston cards.

Thank you all again for your time!

103.508327] mmcblk2: mmc2:0001 00000 59.5 GiB (ro)

As soon as you plug in the card, the kernel sees it as a read-only device.

I don't have any other ideas.

I did a little research on this issue and found some revealing info. Apparently the SD card readers do have the read-only circuitry that it expects to be controlled by the BIOS. Here is some discussion of the problem as it relates to EFI grub: https://www.udoo.org/forum/threads/sd-c ... 699/page-2

At any rate, it is a common hardware issue. Apparently Windows can control this feature.

So what is needed is a linux hack to toggle that RO bit on the SD controller.

[Edit] OR .. try this maybe? https://michael-peeters.blogspot.com/20 ... er-in.html

Ha , blow the bugger out with compressed air. I like that . Hate HP machines of any kind . printers made to hate Linux so are the computers I guess . Same managerial thought processes.

Thank you @ rcrsn51 and @Jafadmin, I guess it's sort of a dead end, I really know nothing about kernel compiling and this type of low level operations. Or maybe a command issued before booting via GRUB? I know some tweakings can be done, but I'm really no expert in this.

About blowing in the sd-card slot (it shouldn't be the case because I can read and write correctly on windows) I've tried unfortunately with no positive results.

I'll try looking at more feasible options.

What are your experiences performance wise with puppy using a cheap thumbdrive as the location for savefile/save directory?
Are the rw operations on the usb a visible bottleneck? How is the responsiveness in case of long sessions?

Maybe after some time and with enough confidence I can then obliterate to oblivion the windows partition and fully install puppy on the internal SSD

pupp_overflow wrote: Fri Mar 05, 2021 9:35 am

About blowing in the sd-card slot (it shouldn't be the case because I can read and write correctly on windows) I've tried unfortunately with no positive results.

That's too bad.

Card readers have always been a problem in Linux. Some work perfectly, some are invisible, some try to work but fail.

Random googling will find various suggestions, but none have ever worked for me.

In my notes, I found a comment about getting a reader to work with a kernel 5.6 setup, but it hasn't worked since.

@pupp_overflow: This doesn't help your situation, but it's an example of how flaky card reader hardware can be:

I finally got the k5.6 machine to start working again by booting it with an Ethernet cable plugged in! I'm guessing that this works because the Ethernet adapter and SD reader are on the same Broadcom subsystem, so the former is needed to initialize the latter. Now I can also use it with newer 5.x kernels.

Unfortunately, that doesn't help on a different machine where an SD card is not detected at all.

rcrsn51 wrote: Fri Mar 05, 2021 4:49 pm

@pupp_overflow: This doesn't help your situation, but it's an example of how flaky card reader hardware can be:

I finally got the k5.6 machine to start working again by booting it with an Ethernet cable plugged in! I'm guessing that this works because the Ethernet adapter and SD reader are on the same Broadcom subsystem, so the former is needed to initialize the latter. Now I can also use it with newer 5.x kernels.

Unfortunately, that doesn't help on a different machine where an SD card is not detected at all.

A timely thread, because just this morning I was saving some document files to SD and I started getting the 'read-only file system' message. Eventually the card just disappeared in pmount. I had to remove it and plug it back in. It reminded me that a month ago I had a similar issue in Fossapup on another computer, which resulted in me corrupting the partition somehow. I re-formatted the card and went on my merry way.

It's a shame, because I recently started using micro SD cards as bulk storage when I realized just how fast, high capacity, small, and cheap they are. Now it seems they may not be as reliable as straight USB storage.

pupp_overflow wrote: Fri Mar 05, 2021 9:35 am

What are your experiences performance wise with puppy using a cheap thumbdrive as the location for savefile/save directory?
Are the rw operations on the usb a visible bottleneck? How is the responsiveness in case of long sessions?

I personally run entire puppy OS's from these little buggers: https://www.amazon.com/SanDisk-256GB-Ul ... 5LJ99?th=1

They are very fast and work like a charm. I have several that I have configured as bootable NAS devices.

Puppy was designed to run as a USB Thumbdrive OS.

Jafadmin wrote: Fri Mar 05, 2021 7:02 pm

pupp_overflow wrote: Fri Mar 05, 2021 9:35 am

What are your experiences performance wise with puppy using a cheap thumbdrive as the location for savefile/save directory?
Are the rw operations on the usb a visible bottleneck? How is the responsiveness in case of long sessions?

I personally run entire puppy OS's from these little buggers: https://www.amazon.com/SanDisk-256GB-Ul ... 5LJ99?th=1

They are very fast and work like a charm. I have several that I have configured as bootable NAS devices.

Puppy was designed to run as a USB Thumbdrive OS.

I also ran Fossapup from an SD card, but I couldn't figure out how to load the save from the same SD card, even though it's the home directory. I wonder if I used the 'save to another partition' option in the frugal installer and chose that drive, would it see the file on bootup then?

geo_c wrote: Fri Mar 05, 2021 7:07 pm

I also ran Fossapup from an SD card, ...

I don't understand what you mean by this. We need to know what your boot device is. Are you saying that you booted from an SD card? If so, was it from an internal SD reader, of from a USB SD reader?

Jafadmin wrote: Sat Mar 06, 2021 1:08 am

geo_c wrote: Fri Mar 05, 2021 7:07 pm

I also ran Fossapup from an SD card, ...

I don't understand what you mean by this. We need to know what your boot device is. Are you saying that you booted from an SD card? If so, was it from an internal SD reader, of from a USB SD reader?

Oh sorry, it's an SD card slot on a laptop, a micro SD card in an adapter, not an internal drive, and I can't boot from it because the bios doesn't seem capable of booting from the SD card. So I installed fossa on the SD card and ran grub4dos on a USB thumb drive to boot the system on the SD card. But I couldn't figure out how to get the fossa save file on the SD card to show up after boot.

rcrsn51 wrote: Fri Mar 05, 2021 4:49 pm

@pupp_overflow: This doesn't help your situation, but it's an example of how flaky card reader hardware can be:

I finally got the k5.6 machine to start working again by booting it with an Ethernet cable plugged in! I'm guessing that this works because the Ethernet adapter and SD reader are on the same Broadcom subsystem, so the former is needed to initialize the latter. Now I can also use it with newer 5.x kernels.

Unfortunately, that doesn't help on a different machine where an SD card is not detected at all.

@rcrsn51 Thank you for your time with extended tests, I tried dabbling with plugs in the micro usb and mini-hdmi ports next to the sd-card slot, but unfortunately nothing worked. At this point the only solution is to just beat it or find a trick in the boot GRUB command I guess.

Jafadmin wrote: Fri Mar 05, 2021 7:02 pm

I personally run entire puppy OS's from these little buggers: https://www.amazon.com/SanDisk-256GB-Ul ... 5LJ99?th=1

They are very fast and work like a charm. I have several that I have configured as bootable NAS devices.

Puppy was designed to run as a USB Thumbdrive OS.

Interesting @Jafadmin thank you for your input, aren't usually small flash drives like these quick to get heated? I have only a single usb 2 port available, would that result in slow execution of the OS during a long session usage?
Can I ask you to explain briefly how you configured those flash drives as bootable NAS? Sounds like an interesting way to reuse a few thumb drives I have lying around.

In any case as of now I think I will test more deeply with FatDog64 since it feels quicker and more snappy with this tablet.

Thank you all for your suggestions and help!

Something to consider: get a little USB hub that would let you plug in multiple USB devices.

After long search the solution for this is a boot parameter! (maybe this works with other devices too?)
"sdhci.debug_quirks=0x10000"
Do you have any considerations about some unexpected side effects with this param on?
Thanks and credits goes to apinela!

Because this is a specific computer make/model hardware issue.
Could you add to the first post subject, the computer make and model.

This would help others trying to use this exact model computer.

Apparently with a quick search for debug_quirks I've found that it's rather a common issue with various laptops from different makers. Funny enough this looks like quite an arcane config, I don't know if this is correct but even in kernel source code the reasons are quite "cryptic" (mind, I'm no expert in these kernel things, so maybe I've found something that it's not that relevant)
https://github.com/torvalds/linux/blob/ ... 4848-L4849

Small excerpt below, note last lines

/*****************************************************************************\
 *                                                                           *
 * Driver init/exit                                                          *
 *                                                                           *
\*****************************************************************************/

static int __init sdhci_drv_init(void)
{
	pr_info(DRIVER_NAME
		": Secure Digital Host Controller Interface driver\n");
	pr_info(DRIVER_NAME ": Copyright(c) Pierre Ossman\n");

	return 0;
}

static void __exit sdhci_drv_exit(void)
{
}

module_init(sdhci_drv_init);
module_exit(sdhci_drv_exit);

module_param(debug_quirks, uint, 0444);
module_param(debug_quirks2, uint, 0444);

MODULE_AUTHOR("Pierre Ossman <pierre@ossman.eu>");
MODULE_DESCRIPTION("Secure Digital Host Controller Interface core driver");
MODULE_LICENSE("GPL");

MODULE_PARM_DESC(debug_quirks, "Force certain quirks.");
MODULE_PARM_DESC(debug_quirks2, "Force certain other quirks.");

geo_c wrote: Fri Mar 05, 2021 7:02 pm

rcrsn51 wrote: Fri Mar 05, 2021 4:49 pm

@pupp_overflow: This doesn't help your situation, but it's an example of how flaky card reader hardware can be:

I finally got the k5.6 machine to start working again by booting it with an Ethernet cable plugged in! I'm guessing that this works because the Ethernet adapter and SD reader are on the same Broadcom subsystem, so the former is needed to initialize the latter. Now I can also use it with newer 5.x kernels.

Unfortunately, that doesn't help on a different machine where an SD card is not detected at all.

A timely thread, because just this morning I was saving some document files to SD and I started getting the 'read-only file system' message. Eventually the card just disappeared in pmount. I had to remove it and plug it back in. It reminded me that a month ago I had a similar issue in Fossapup on another computer, which resulted in me corrupting the partition somehow. I re-formatted the card and went on my merry way.

It's a shame, because I recently started using micro SD cards as bulk storage when I realized just how fast, high capacity, small, and cheap they are. Now it seems they may not be as reliable as straight USB storage.

@geo_c :-

Don't forget, there's a few companies out there now making these teeny weeny 'nano'-sized flash drives (SanDisk & Integral, to mention but two). And SanDisk do a 512 GB (half a terabyte) version of their USB 3.1 Gen1 variant:-

https://www.amazon.co.uk/SanDisk-Ultra- ... r=1-3&th=1

.....and so does Integral:-

https://www.amazon.co.uk/Integral-Fusio ... r=1-3&th=1

Just FYI.

Mike.