Retry for sdio write errors.

[nsloss]

Hi,

I've updated to the latest 2024-10-26 release
And adding clock delay
[SDIO]
AddClockDelay=1

Still resulted in errors in the log.txt

Platform: BlueSCSI
FW Version: 2024.10.26-rel Oct 26 2024 22:02:38
I2C Supported
Flash chip size: 2048 kB
INFO: Injecting 1 additional wait state(s) on SDIO
 
=== SD Card Info ===
SD card detected, exFAT volume size: 61088 MB
SD Name: EC1S5, MID: 0x1B, OID: 0x53 0x4D
 
=== Global Config ===
Reading configuration from bluescsi.ini
Active configuration:
 
=== Finding images in / ===
== Opening /HD00_512.hda for ID: 0 LUN: 0
---- Image ready
== Opening /NE4.hda for ID: 4 LUN: 0
---- Configuring as network based on image name
---- Image ready

=== ROM Drive ===
Platform supports ROM drive up to 1692 kB
---- ROM drive image not detected
 
=== Configured SCSI Devices ===
* ID: 0, BlockSize: 512, Type: Fixed, Quirks: Apple, Size: 20480000kB
* ID: 4, Type: Network, Quirks: Apple
 
=== Network Initialization ===
Wi-Fi MAC: xx:xx:xx:xx:xx:xx
Connecting to Wi-Fi SSID "MYSSID" with WPA/WPA2 PSK
 
Initialization complete!
INFO: Pico Voltage: 3.357V.
Successfully connected to Wi-Fi SSID "NATMAN"
SDIO card reports write CRC error, status 0x000000EB
SdioCard::writeSectors(0x0075F2A0,...,16) failed: 7
SD card write failed: 0x00

Retrying on SDIO WRITE ERRORS was successful after applying the following patch:

--- ./lib/BlueSCSI_platform_RP2040/rp2040_sdio.cpp      2024-10-27 05:41:00.000000000 +1100
+++ /home/nat/rp2040_sdio.cpp   2024-11-20 18:58:31.423144365 +1100
@@ -56,6 +56,7 @@ static struct {
     uint32_t end_token_buf[3]; // CRC and end token for write block
     sdio_status_t wr_status;
     uint32_t card_response;
+    uint8_t retries;
 
     // Variables for block reads
     // This is used to perform DMA into data buffers and checksum buffers separately.
@@ -706,6 +707,8 @@ sdio_status_t rp2040_sdio_tx_start(const
     g_sdio.total_blocks = num_blocks;
     g_sdio.blocks_checksumed = 0;
     g_sdio.checksum_errors = 0;
+    g_sdio.retries = 0;
+    g_sdio.wr_status = SDIO_OK;
 
     // Compute first block checksum
     sdio_compute_next_tx_checksum();
@@ -713,12 +716,6 @@ sdio_status_t rp2040_sdio_tx_start(const
     // Start first DMA transfer and PIO
     sdio_start_next_block_tx();
 
-    if (g_sdio.blocks_checksumed < g_sdio.total_blocks)
-    {
-        // Precompute second block checksum
-        sdio_compute_next_tx_checksum();
-    }
-
     return SDIO_OK;
 }
 
@@ -788,10 +785,25 @@ static void rp2040_sdio_tx_irq()
             g_sdio.wr_status = check_sdio_write_response(g_sdio.card_response);
 
             if (g_sdio.wr_status != SDIO_OK)
-            {
-                rp2040_sdio_stop();
-                return;
-            }
+           {
+               g_sdio.retries++;
+               if (g_sdio.retries > 3)
+               {
+                    rp2040_sdio_stop();
+                    return;
+               }
+               else if (g_sdio.blocks_done < g_sdio.total_blocks)
+               {
+                   g_sdio.wr_status = SDIO_OK;
+                    sdio_start_next_block_tx();
+                    g_sdio.transfer_state = SDIO_TX;
+                   return;
+               }
+           }
+
+           g_sdio.retries = 0;
+            if (g_sdio.blocks_checksumed < g_sdio.total_blocks)
+                sdio_compute_next_tx_checksum();
 
             g_sdio.blocks_done++;
             if (g_sdio.blocks_done < g_sdio.total_blocks)
@@ -799,12 +811,6 @@ static void rp2040_sdio_tx_irq()
                 sdio_start_next_block_tx();
                 g_sdio.transfer_state = SDIO_TX;
 
-                if (g_sdio.blocks_checksumed < g_sdio.total_blocks)
-                {
-                    // Precompute the CRC for next block so that it is ready when
-                    // we want to send it.
-                    sdio_compute_next_tx_checksum();
-                }
             }
             else
             {

This will retry writing the current block up to 3 times and so far this has been successful reporting no errors on NetBSD-current

BlueSCSI-v2 with wifi Powebook Edition.
20G Image size.

I hope this helps,

Best regards,

Nat

[nsloss]

The patch was garbled by the web ui so I opened pull request:
#211

With the patch in full.

Best regards,

Nat

[nsloss]

Sorry acidently clicked on close it should remain open until the pull request has been handled

[nsloss]

reopening... my apologies im a beginner

[erichelgeson]

Thanks for taking the time to dig in and no need to apologize!

I see you tested with the 10/26 release - Is there anyway you could re-test with the latest nightly (without your patch) and see if you experience the same result? We just merged #209 which should help with this case, but it'd be great to know if it did/didnt.

Also you say you're running NetBSD on your powerbook? Cool! I thought the NetBSD Wi-Fi driver wasn't compatible or was there an update? Either way any updates to the wiki with any tips or tricks (or prebuilt base image) would be appreciated.

[erichelgeson]

Confirmed resolved in #209