I can provide some related context that explains the behavior you observed.
Modern operating systems do their own caching on the OS side. applications write to a cache in-memory which is then written to the device itself when it becomes available. This allows applications to stay responsive, but can also create the appearance that data has been written even when it has not. If for some reason a drive is not writing AND is not reporting an error, you will see the change until the drive is remounted. This was a common "problem" with SATA writeblockers I supported in the past. Customers would call saying the data was still being written even though the blocker was being used. I would then have to explain to them how the OS caching works and demonstrate the effect. This explains the behavior you saw, but doesn't answer why. Unfortunately the challenge you run into is that the write failure does not trigger an error that Linux can see, resulting in the OS believing the write to have succeeded. Or maybe these are old SD cards that have simply locked themselves into RO mode. -Ben On Sunday, May 10th, 2026 at 8:16 PM, Russell Senior <[email protected]> wrote: > I like to think I am open to contradiction. I tell people that I like > when they tell me I'm wrong about something, particularly if they can > convincingly explain why. > > -- > Russell > > On Sun, May 10, 2026 at 6:06 PM Tomas Kuchta > <[email protected]> wrote: > > > > No need to counter speculate - Russell has confirmed that my situation was > > not matching his test/data. > > > > Russell is authoritative souce of reality for these SD cards. > > > > :-) T > > > > On Sun, May 10, 2026, 15:26 wes <[email protected]> wrote: > > > > > I counter-speculate that this is a failure mode of flash storage: the > > > controller onboard the storage device is able to accept writes into its > > > cache, but fails to flush the cache to the actual storage media. > > > > > > -wes > > > > > > On Sun, May 10, 2026 at 6:09 AM Tomas Kuchta <[email protected] > > > > > > > wrote: > > > > > > > I have experienced this before and not really understood it, due to the > > > > lack of patience and finding a workaround. > > > > > > > > My workaround was to use the Raspberry Pi imager writing what I needed > > > > to > > > > the SD card. I believe, if my memory serves me well, that Balena Etcher > > > was > > > > also successful at writing to these SD cards. Both imagers run on Linux. > > > > > > > > From that - I guess that you can repartition the SD then write normal > > > data > > > > from /dev/zero to the partitions. > > > > > > > > My wild guess was that the SD card firmware will block if trying to > > > > write > > > > outside the partitions boundaries. Perhaps to protect itself from being > > > > owerwritten. That is my speculation though. > > > > > > > > -T > > > > > > > > On Sat, May 9, 2026, 23:47 Russell Senior <[email protected]> > > > > wrote: > > > > > > > > > On Sat, May 9, 2026 at 3:56 PM Ted Mittelstaedt < > > > [email protected]> > > > > > wrote: > > > > > > > > > > > > Have you tried a different card reader in a different computer? > > > > > > > > > > I think I literally said that in the second paragraph, but yes. On > > > > > plugging via a USB dongle, dmesg says it is writable: > > > > > > > > > > [Sat May 9 20:06:58 2026] usb 3-2: new high-speed USB device number > > > > > 18 using xhci_hcd > > > > > [Sat May 9 20:06:58 2026] usb 3-2: New USB device found, > > > > > idVendor=05e3, idProduct=0751, bcdDevice=14.04 > > > > > [Sat May 9 20:06:58 2026] usb 3-2: New USB device strings: Mfr=3, > > > > > Product=4, SerialNumber=0 > > > > > [Sat May 9 20:06:58 2026] usb 3-2: Product: USB Storage > > > > > [Sat May 9 20:06:58 2026] usb 3-2: Manufacturer: USB Storage > > > > > [Sat May 9 20:06:58 2026] usb-storage 3-2:1.0: USB Mass Storage > > > > > device > > > > > detected > > > > > [Sat May 9 20:06:58 2026] scsi host0: usb-storage 3-2:1.0 > > > > > [Sat May 9 20:06:59 2026] scsi 0:0:0:0: Direct-Access Generic > > > > > STORAGE DEVICE 1404 PQ: 0 ANSI: 6 > > > > > [Sat May 9 20:06:59 2026] sd 0:0:0:0: Attached scsi generic sg0 type > > > > > 0 > > > > > [Sat May 9 20:06:59 2026] sd 0:0:0:0: [sda] 31116288 512-byte logical > > > > > blocks: (15.9 GB/14.8 GiB) > > > > > [Sat May 9 20:06:59 2026] sd 0:0:0:0: [sda] Write Protect is off > > > > > [Sat May 9 20:06:59 2026] sd 0:0:0:0: [sda] Mode Sense: 21 00 00 00 > > > > > [Sat May 9 20:06:59 2026] sd 0:0:0:0: [sda] Write cache: disabled, > > > > > read cache: enabled, doesn't support DPO or FUA > > > > > [Sat May 9 20:06:59 2026] sda: sda1 > > > > > [Sat May 9 20:06:59 2026] sd 0:0:0:0: [sda] Attached SCSI removable > > > disk > > > > > [Sat May 9 20:07:00 2026] EXT4-fs (sda1): recovery complete > > > > > [Sat May 9 20:07:00 2026] EXT4-fs (sda1): mounted filesystem > > > > > 40af6f66-8d84-45b4-b056-9a2ee4c5a7b5 r/w with ordered data mode. Quota > > > > > mode: none. > > > > > > > > > > umount /dev/sda1 > > > > > > > > > > [Sat May 9 20:08:02 2026] EXT4-fs (sda1): unmounting filesystem > > > > > 40af6f66-8d84-45b4-b056-9a2ee4c5a7b5. > > > > > > > > > > # time dd if=/dev/urandom of=/dev/sda bs=128k status=progress > > > oflag=sync > > > > > 15921446912 bytes (16 GB, 15 GiB) copied, 863 s, 18.4 MB/s > > > > > dd: error writing '/dev/sda': No space left on device > > > > > 121549+0 records in > > > > > 121548+0 records out > > > > > 15931539456 bytes (16 GB, 15 GiB) copied, 863.936 s, 18.4 MB/s > > > > > > > > > > real 14m23.938s > > > > > user 0m0.749s > > > > > sys 2m5.471s > > > > > > > > > > [Sat May 9 20:23:31 2026] sda: sda1 > > > > > > > > > > like, why is there still a partition? > > > > > > > > > > # time dd if=/dev/zero of=/dev/sda bs=128k status=progress oflag=sync > > > > > 15915155456 bytes (16 GB, 15 GiB) copied, 751 s, 21.2 MB/s > > > > > dd: error writing '/dev/sda': No space left on device > > > > > 121549+0 records in > > > > > 121548+0 records out > > > > > 15931539456 bytes (16 GB, 15 GiB) copied, 752.161 s, 21.2 MB/s > > > > > > > > > > real 12m32.163s > > > > > user 0m0.634s > > > > > sys 0m19.809s > > > > > > > > > > [Sat May 9 20:36:49 2026] sda: sda1 > > > > > > > > > > Unplug and replug, dmesg says: > > > > > > > > > > [Sat May 9 20:39:21 2026] usb 3-2: USB disconnect, device number 18 > > > > > [Sat May 9 20:39:25 2026] usb 3-2: new high-speed USB device number > > > > > 19 using xhci_hcd > > > > > [Sat May 9 20:39:25 2026] usb 3-2: New USB device found, > > > > > idVendor=05e3, idProduct=0751, bcdDevice=14.04 > > > > > [Sat May 9 20:39:25 2026] usb 3-2: New USB device strings: Mfr=3, > > > > > Product=4, SerialNumber=0 > > > > > [Sat May 9 20:39:25 2026] usb 3-2: Product: USB Storage > > > > > [Sat May 9 20:39:25 2026] usb 3-2: Manufacturer: USB Storage > > > > > [Sat May 9 20:39:25 2026] usb-storage 3-2:1.0: USB Mass Storage > > > > > device > > > > > detected > > > > > [Sat May 9 20:39:25 2026] scsi host0: usb-storage 3-2:1.0 > > > > > [Sat May 9 20:39:26 2026] scsi 0:0:0:0: Direct-Access Generic > > > > > STORAGE DEVICE 1404 PQ: 0 ANSI: 6 > > > > > [Sat May 9 20:39:26 2026] sd 0:0:0:0: Attached scsi generic sg0 type > > > > > 0 > > > > > [Sat May 9 20:39:27 2026] sd 0:0:0:0: [sda] 31116288 512-byte logical > > > > > blocks: (15.9 GB/14.8 GiB) > > > > > [Sat May 9 20:39:27 2026] sd 0:0:0:0: [sda] Write Protect is off > > > > > [Sat May 9 20:39:27 2026] sd 0:0:0:0: [sda] Mode Sense: 21 00 00 00 > > > > > [Sat May 9 20:39:27 2026] sd 0:0:0:0: [sda] Write cache: disabled, > > > > > read cache: enabled, doesn't support DPO or FUA > > > > > [Sat May 9 20:39:27 2026] sda: sda1 > > > > > [Sat May 9 20:39:27 2026] sd 0:0:0:0: [sda] Attached SCSI removable > > > disk > > > > > [Sat May 9 20:39:27 2026] EXT4-fs (sda1): recovery complete > > > > > [Sat May 9 20:39:27 2026] EXT4-fs (sda1): mounted filesystem > > > > > 40af6f66-8d84-45b4-b056-9a2ee4c5a7b5 r/w with ordered data mode. Quota > > > > > mode: none. > > > > > > > > > > wtf? > > > > > > > > > > Same thing seems to be happening on another micro SD card on an mmcblk > > > > > interface on a different computer. > > > > > > > > > > > > > > > Sent: Saturday, May 9, 2026 2:15 PM > > > > > > To: Portland Linux/Unix Group <[email protected]> > > > > > > Subject: [PLUG] writing to micro SD card and not writing to micro SD > > > > > cards AT THE SAME TIME??? > > > > > > > > > > > > I was trying to reformat some microSD cards last night. They are > > > > > apparently writable, mount rw. I umount them, use (CAREFULLY) dd to > > > write > > > > > /dev/zero or /dev/urandom on to them and dd seems to happily do it. I > > > can > > > > > even hexdump back out the zeros or whatever back out, for a short > > > > > time. > > > > But > > > > > moments later, all the original content is still there. > > > > > > > > > > > > I have tried multiple computers, two different microSD cards, > > > multiple > > > > > distribution versions, interface adapters, including your usual USB > > > > > adapters and mmcblk adapters, same result. It isn't a write protect > > > slide > > > > > switch, because there aren't any in some of those adapters and anyway, > > > in > > > > > the case where there was one, it was in the correct position. > > > > > > > > > > > > Anyone know what's going on? > > > > > > > > > > > > -- > > > > > > Russell Senior > > > > > > [email protected] > > > > > > > > > > > > > > > > > > >
