target/linux/generic/pending-6.1/803-nvmem-core-fix-support-for-fixed-cells-NVMEM-layout.patch

   1 From: =?UTF-8?q?Rafa=C5=82=20Mi=C5=82ecki?= <rafal@milecki.pl>
   2 Date: Thu, 13 Jul 2023 17:30:59 +0200
   3 Subject: [PATCH] nvmem: core: fix support for fixed cells NVMEM layout
   4 MIME-Version: 1.0
   5 Content-Type: text/plain; charset=UTF-8
   6 Content-Transfer-Encoding: 8bit
   7
   8 Returning -EPROBE_DEFER for "fixed-layout" makes nvmem_register() always
   9 fail (that layout is supported internally with no external module). That
  10 makes callers (e.g. mtd_nvmem_add()) fail as well and prevents booting
  11 on devices with "fixed-layout" in DT.
  12
  13 Add a quick workaround for it.
  14
  15 Fixes: 6468a6f45148 ("nvmem: core: handle the absence of expected layouts")
  16 Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
  17 ---
  18
  19 --- a/drivers/nvmem/core.c
  20 +++ b/drivers/nvmem/core.c
  21 @@ -794,6 +794,19 @@ static struct nvmem_layout *nvmem_layout
  22                 return NULL;
  23
  24         /*
  25 +        * We should return -EPROBE_DEFER only when layout driver is expected to
  26 +        * become available later. Otherwise NVMEM will never probe successfully
  27 +        * for unsupported layouts. There is no known solution for that right
  28 +        * now.
  29 +        *
  30 +        * This problem also affects "fixed-layout". It's supported in NVMEM
  31 +        * core code so there never will be layout for it. We shouldn't return
  32 +        * -EPROBE_DEFER in such case. Add a quick workaround for that.
  33 +        */
  34 +       if (of_device_is_compatible(layout_np, "fixed-layout"))
  35 +               return NULL;
  36 +
  37 +       /*
  38          * In case the nvmem device was built-in while the layout was built as a
  39          * module, we shall manually request the layout driver loading otherwise
  40          * we'll never have any match.