The ucode chunk might be relatively large and the allocation with
kmalloc() may fail occasionally. Since the data isn't DMA-transferred
but by manual loops, we can use vmalloc instead of kmalloc.
For a better performance, though, kvmalloc() would be the best choice
in such a case, so let's replace with it.
Bugzilla: https://bugzilla.suse.com/show_bug.cgi?id=
1103431
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
if (le32_to_cpu(hdr->idx) == idx) {
pdata = wl->fw.fw_bin[i]->data +
le32_to_cpu(hdr->offset);
- *pbuf = kmemdup(pdata, len, GFP_KERNEL);
+ *pbuf = kvmalloc(len, GFP_KERNEL);
if (*pbuf == NULL)
goto fail;
-
+ memcpy(*pbuf, pdata, len);
return 0;
}
}
*/
void brcms_ucode_free_buf(void *p)
{
- kfree(p);
+ kvfree(p);
}
/*