ipq4019 kernel 报错spi-nand spi0.1: spi transfer failed: -110

原因是：spi nand flash有坏快，可以合入如下的patch，提高spi的性能tiao过坏块

报错：SQUASHFS error: Unable to read page, block 325e36, size c97

合入如下patch：

提高了 SQUASHFS 处理读取错误的能力。经过应用此补丁，缓存中损坏的页面将不再导致至少在大多数情况下，减压持续失败。

为 SPI QUP 驱动程序实现 `handle_err` 函数。此外，删除呼叫到两个已弃用的内核 API 函数，将它们替换为当前的推荐的版本。

9996-fs_squashfs_improve_squashfs_error_resistance.patch

From: Oever González <notengobattery@gmail.com>
Subject: [PATCH] fs: SQUASHFS: improve SQUASHFS error resistance
Date: Sat, 25 Jul 2020 18:49:31 -0600This patch greatly improves the SQUASHFS's ability to deal with read errors. By
applying this patch, a corrupted page in the cache will not longer lead to a
persistent failure in decompression, at least most of the times.Signed-off-by: Oever González <notengobattery@gmail.com>
---
--- a/fs/squashfs/block.c
+++ b/fs/squashfs/block.c
@@ -65,6 +65,42 @@ static struct buffer_head *get_block_length(struct super_block *sb,return bh;}+/*
+ * Kill the pages if an error is found, this will try to evict them from the
+ * page cache forcing the kernel to read again the device, hopping that the read
+ * error is just transitory.
+ *
+ * This efectively cleans the "back" cache. Normal file systems will do this,
+ * since this makes them more resistant to sparse or transitory I/O failures.
+ *
+ */
+static inline void __squashfs_kill_pages(struct buffer_head **bh, int quantity)
+{
+  int index = 0, total_pages = 0;
+  struct buffer_head *bh_head;
+  struct buffer_head *bh_current;
+
+  for (index = 0; index < quantity; index++) {
+      bh_head = bh[index];
+      bh_current = bh[index];
+
+      do {
+          struct page *bh_page = bh_current->b_page;
+
+          lock_page(bh_page);
+          delete_from_page_cache(bh_page);
+          ClearPageUptodate(bh_page);
+          SetPageError(bh_page);
+          total_pages++;
+          unlock_page(bh_page);
+
+          clear_buffer_uptodate(bh_current);
+          bh_current = bh_current->b_this_page;
+      } while (bh_current != bh_head);
+  }
+
+  WARNING("killed %d pages, %d buffer heads\n", total_pages, quantity);
+}/** Read and decompress a metadata block or datablock.  Length is non-zero
@@ -75,8 +111,8 @@ static struct buffer_head *get_block_length(struct super_block *sb,* generated a larger block - this does occasionally happen with compression* algorithms).*/
-int squashfs_read_data(struct super_block *sb, u64 index, int length,
-       u64 *next_index, struct squashfs_page_actor *output)
+static inline int __squashfs_read_data(struct super_block *sb, u64 index,
+      int length, u64 *next_index, struct squashfs_page_actor *output){struct squashfs_sb_info *msblk = sb->s_fs_info;struct buffer_head **bh;
@@ -194,11 +230,51 @@ int squashfs_read_data(struct super_block *sb, u64 index, int length,block_release:for (; k < b; k++)
-       put_bh(bh[k]);
+      brelse(bh[k]);read_failure:
-   ERROR("squashfs_read_data failed to read block 0x%llx\n",
-                   (unsigned long long) index);
+  __squashfs_kill_pages(bh, b);kfree(bh);return -EIO;}
+
+/*
+ * If some kind of error is detected, block into this loop rather than crashing
+ * the process that requested the data, since it can be `init` and crashing it
+ * will lead to a kernel panic. If the read still failing, the process is doomed
+ * to crash anyway.
+ *
+ * This only makes SQUASHFS more error resistant by avoiding the poisoning of
+ * the "front" cache if the first attempt failed.
+ *
+ */
+int squashfs_read_data(struct super_block *sb, u64 index, int length,
+      u64 *next_index, struct squashfs_page_actor *output)
+{
+  int ret = 0, attempts = 0;
+  unsigned long long block = (unsigned long long) index;
+  u64 saved_next_index;
+
+  if (next_index)
+      saved_next_index = *next_index;
+
+  ret = __squashfs_read_data(sb, index, length, next_index, output);
+
+  while (ret < 0 && attempts < 5) {  // Retry 5 times, a total of 6 attempts
+      attempts++;
+      TRACE("failed to read block [%llx], retry attempt %d\n",
+          block, attempts);
+      if (next_index)
+          *next_index = saved_next_index;
+      ret = __squashfs_read_data(sb, index, length, next_index, output);
+  }
+
+  if (attempts > 0 && ret >= 0)
+      TRACE("read_data: success after %d attempts to read block [%llx]\n",
+           attempts, block);
+  else if (attempts > 0 && ret < 0)
+      ERROR("read_data: failed after %d attempts to read block [%llx]\n",
+            attempts + 1);
+
+  return ret;
+}
--- a/fs/squashfs/cache.c
+++ b/fs/squashfs/cache.c
@@ -112,8 +112,12 @@ struct squashfs_cache_entry *squashfs_cache_get(struct super_block *sb,spin_lock(&cache->lock);-           if (entry->length < 0)
+          if (entry->length < 0) {entry->error = entry->length;
+              WARNING("Invalidated %s cache entry [%llx]\n", cache->name,
+                  entry->block);
+              entry->block = SQUASHFS_INVALID_BLK;
+          }entry->pending = 0;--- a/fs/squashfs/decompressor_multi.c
+++ b/fs/squashfs/decompressor_multi.c
@@ -189,8 +189,6 @@ int squashfs_decompress(struct squashfs_sb_info *msblk, struct buffer_head **bh,res = msblk->decompressor->decompress(msblk, decomp_stream->stream,bh, b, offset, length, output);put_decomp_stream(decomp_stream, stream);
-   if (res < 0)
-       ERROR("%s decompression failed, data probably corrupt\n",
-           msblk->decompressor->name);
+return res;}
--- a/fs/squashfs/decompressor_multi_percpu.c
+++ b/fs/squashfs/decompressor_multi_percpu.c
@@ -82,10 +82,6 @@ int squashfs_decompress(struct squashfs_sb_info *msblk, struct buffer_head **bh,offset, length, output);put_cpu_ptr(stream);-   if (res < 0)
-       ERROR("%s decompression failed, data probably corrupt\n",
-           msblk->decompressor->name);
-return res;}--- a/fs/squashfs/decompressor_single.c
+++ b/fs/squashfs/decompressor_single.c
@@ -70,10 +70,6 @@ int squashfs_decompress(struct squashfs_sb_info *msblk, struct buffer_head **bh,offset, length, output);mutex_unlock(&stream->mutex);-   if (res < 0)
-       ERROR("%s decompression failed, data probably corrupt\n",
-           msblk->decompressor->name);
-return res;}--- a/fs/squashfs/super.c
+++ b/fs/squashfs/super.c
@@ -452,6 +452,7 @@ static int __init init_squashfs_fs(void)}pr_info("version 4.0 (2009/01/31) Phillip Lougher\n");
+  pr_info("patched 4.0 (2020/11/10) Oever Gonzalez\n");return 0;}

9994-mtd_ubi_improve_the_i_o_retries.patch

From: Oever González <notengobattery@gmail.com>
Subject: [PATCH] mtd: UBI: improve the I/O retries
Date: Thu, 23 Jul 2020 09:41:17 -0600In order to make UBI more error resistant, and because this also affects the
stress test (to check if a PEB has died), improve the retries number from the
current value of 3 to 4.Signed-off-by: Oever González <notengobattery@gmail.com>
---
--- a/drivers/mtd/ubi/io.c
+++ b/drivers/mtd/ubi/io.c
@@ -76,6 +76,7 @@#include <linux/crc32.h>#include <linux/err.h>#include <linux/slab.h>
+#include <linux/delay.h>#include "ubi.h"static int self_check_not_bad(const struct ubi_device *ubi, int pnum);
@@ -173,7 +174,7 @@ int ubi_io_read(const struct ubi_device *ubi, void *buf, int pnum, int offset,if (retries++ < UBI_IO_RETRIES) {ubi_warn(ubi, "error %d%s while reading %d bytes from PEB %d:%d, read only %zd bytes, retry",err, errstr, len, pnum, offset, read);
-           yield();
+          msleep(100);goto retry;}@@ -328,7 +329,7 @@ static int do_sync_erase(struct ubi_device *ubi, int pnum)if (retries++ < UBI_IO_RETRIES) {ubi_warn(ubi, "error %d while erasing PEB %d, retry",err, pnum);
-           yield();
+          msleep(100);goto retry;}ubi_err(ubi, "cannot erase PEB %d, error %d", pnum, err);
--- a/drivers/mtd/ubi/ubi.h
+++ b/drivers/mtd/ubi/ubi.h
@@ -63,7 +63,7 @@ void ubi_err(const struct ubi_device *ubi, const char *fmt, ...);* In case of errors, UBI tries to repeat the operation several times before* returning error. The below constant defines how many times UBI re-tries.*/
-#define UBI_IO_RETRIES 3
+#define UBI_IO_RETRIES 4/** Length of the protection queue. The length is effectively equivalent to the

9995-spi_qup_implement_error_handle_function.patch

From: Oever González <notengobattery@gmail.com>
Subject: [PATCH] SPI: QUP: implement error handle function
Date: Thu, 23 Jul 2020 22:19:28 -0600Implement the `handle_err` function for the SPI QUP driver. Also, remove calls
to two deprecated kernel API functions, replacing them with the currently
recommended versions.Signed-off-by: Oever González <notengobattery@gmail.com>
---
--- a/drivers/spi/spi-qup.c
+++ a/drivers/spi/spi-qup.c
@@ -417,9 +417,9 @@ static void spi_qup_dma_terminate(struct spi_master *master,struct spi_transfer *xfer){if (xfer->tx_buf)
-       dmaengine_terminate_all(master->dma_tx);
+      dmaengine_terminate_sync(master->dma_tx);if (xfer->rx_buf)
-       dmaengine_terminate_all(master->dma_rx);
+      dmaengine_terminate_sync(master->dma_rx);}static u32 spi_qup_sgl_get_nents_len(struct scatterlist *sgl, u32 max,
@@ -932,11 +932,11 @@ static int spi_qup_init_dma(struct spi_master *master, resource_size_t base)int ret;/* allocate dma resources, if available */
-   master->dma_rx = dma_request_slave_channel_reason(dev, "rx");
+  master->dma_rx = dma_request_chan(dev, "rx");if (IS_ERR(master->dma_rx))return PTR_ERR(master->dma_rx);-    master->dma_tx = dma_request_slave_channel_reason(dev, "tx");
+  master->dma_tx = dma_request_chan(dev, "tx");if (IS_ERR(master->dma_tx)) {ret = PTR_ERR(master->dma_tx);goto err_tx;
@@ -992,6 +992,24 @@ static void spi_qup_set_cs(struct spi_device *spi, bool val)writel_relaxed(spi_ioc, controller->base + SPI_IO_CONTROL);}+static void spi_qup_handle_err(struct spi_master *master,
+               struct spi_message *msg)
+{
+  struct spi_qup *controller = spi_master_get_devdata(master);
+  unsigned long flags;
+
+  spin_lock_irqsave(&controller->lock, flags);
+  controller->error    = 0;
+  controller->rx_bytes = 0;
+  controller->tx_bytes = 0;
+  spin_unlock_irqrestore(&controller->lock, flags);
+
+  spi_qup_set_state(controller, QUP_STATE_RESET);
+
+  if (spi_qup_is_dma_xfer(controller->mode))
+      spi_qup_dma_terminate(master, controller->xfer);
+}
+static int spi_qup_probe(struct platform_device *pdev){struct spi_master *master;
@@ -1063,6 +1081,7 @@ static int spi_qup_probe(struct platform_device *pdev)master->bits_per_word_mask = SPI_BPW_RANGE_MASK(4, 32);master->max_speed_hz = max_freq;master->transfer_one = spi_qup_transfer_one;
+  master->handle_err = spi_qup_handle_err;master->dev.of_node = pdev->dev.of_node;master->auto_runtime_pm = true;master->dma_alignment = dma_get_cache_alignment();

ipq4019 kernel 报错spi-nand spi0.1: spi transfer failed: -110相关推荐

SAP WM初阶LQ02报错 - Movement Type 901 for manual transfer orders does not exist -
SAP WM初阶LQ02报错 - Movement Type 901 for manual transfer orders does not exist - SAP WM模块里的事务代码LQ02很有用 ...
Tensorflow 报错：tensorflow.python.framework.errors_impl.InternalError: Failed to create session....
问题描述 IDE:pycharm,环境中安装tensorflow-gpu 1.8.0 ,Cuda9 ,cudnn 7,等,运行代码报错如下 tensorflow.python.framework.e ...
Ubuntu下Linux系统部署fisco时bash操作报错权限不足（permission denied） failed to run command ‘........‘
Ubuntu下Linux系统部署fisco时bash操作报错权限不足(permission denied) failed to run command '-' 例如这里就指的是fisco-bcos这个 ...
解决win10下 git闪退的问题（或者使用git命令报错fatal: open /dev/null or dup failed）
解决win10下 git闪退的问题(或者使用git命令报错fatal: open /dev/null or dup failed) 本人在win10下安装git完成后,打开闪退,在"运行&q ...
npm run build报错Ineffective mark-compacts near heap limit Allocation failed - JavaScript heap out of
在react项目执行 npm run build 报错 Ineffective mark-compacts near heap limit Allocation failed - JavaScript ...
jupyter notebook导入numpy 失败报错：Original error was: DLL load failed while importing _multiarray_umath:
jupyter notebook导入numpy 失败报错:Original error was: DLL load failed while importing _multiarray_umath: ...
AndroidStudio编译报错org.gradle.api.tasks.TaskExecutionException: Execution failed for task ':app:proces
AndroidStudio编译报错 org.gradle.api.tasks.TaskExecutionException: Execution failed for task ':app:proce ...
SPI NAND Flash和SPI NOR Flash的区别
我们在购买电子产品时,常常听到FLASH闪存这个词.但对于基础小白来说,可能常常搞不清楚SPI Flash.Nand Flash.Nor Flash等都是指什么,今天宏旺半导体就跟大家通俗易懂地讲解一 ...
EASYPOI导入报错The part /xl/drawings/drawing1.xml failed to be saved
EASYPOI导入报错The part /xl/drawings/drawing1.xml failed to be saved Caused by: org.apache.poi.openxml4j ...
MySQL删除用户报错:ERROR 1396 (HY000): Operation DROP USER failed for ‘tsjt‘@‘127.0.0.1‘
背景: 本地安装的MySQL的MariaDB,期间创建了两个名为tsjt的不同登录ip的用户,一个为127.0.0.1,另一个为localhost.测试完毕后准备删除. 此时却发现创建的tsjt@12 ...

ipq4019 kernel 报错spi-nand spi0.1: spi transfer failed: -110

9994-mtd_ubi_improve_the_i_o_retries.patch

9995-spi_qup_implement_error_handle_function.patch

ipq4019 kernel 报错spi-nand spi0.1: spi transfer failed: -110相关推荐

最新文章

热门文章