[PATCH 3/4] nvmem: add a driver for the Amlogic Meson6/Meson8/Meson8b SoCs

On 03/10/17 05:48, Chris Moore wrote:

Hi,

Sorry in advance if I am being extremely stupid here.

Le 01/10/2017 ? 14:56, Martin Blumenstingl a ?crit :

[snip]

quoted

+static int meson_mx_efuse_read(void *context, unsigned int offset,
+                   void *buf, size_t bytes)
+{
+    struct meson_mx_efuse *efuse = context;
+    u32 tmp;
+    int err, i, addr;
+
+    err = meson_mx_efuse_hw_enable(efuse);
+    if (err)
+        return err;
+
+    meson_mx_efuse_mask_bits(efuse, MESON_MX_EFUSE_CNTL1,
+                 MESON_MX_EFUSE_CNTL1_AUTO_RD_ENABLE,
+                 MESON_MX_EFUSE_CNTL1_AUTO_RD_ENABLE);
+
+    for (i = offset; i < offset + bytes; i += efuse->config.word_size) {
+        addr = i / efuse->config.word_size;
+
+        err = meson_mx_efuse_read_addr(efuse, addr, &tmp);
+        if (err)
+            break;
+
+        memcpy(buf + i, &tmp, efuse->config.word_size);
+    }

This will not give the expected result if offset is not a multiple of 
efuse->config.word_size.

Offset should be always multiple of stride which is enforced in nvmem core.

nvmem_cell would return an -EINVAL if that is not the case.


--srini

This will cause buffer overflow if bytes is not a multiple of 
efuse->config.word_size.
Shouldn't there at least be some sort of test on offset and bytes?

Also this doesn't look endian-safe, but maybe it is.

Old-timer remarks:
- I try to avoid divisions where possible but I suppose they are very 
low-cost on recent hardware.
- I don't like arithmetic on void pointers but I know it is supported by 
gcc (and possibly by recent C standards too).

Sorry again for the noise if I am being stupid.

Cheers,
Chris