linux-kernel - Re: [PATCH v1 07/14] nvmem: core: add per-cell post processing

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:   Tue, 30 Aug 2022 16:20:54 +0200
From:   Michael Walle <michael@...le.cc>
To:     Srinivas Kandagatla <srinivas.kandagatla@...aro.org>
Cc:     Miquel Raynal <miquel.raynal@...tlin.com>,
        Richard Weinberger <richard@....at>,
        Vignesh Raghavendra <vigneshr@...com>,
        Rob Herring <robh+dt@...nel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
        Shawn Guo <shawnguo@...nel.org>, Li Yang <leoyang.li@....com>,
        Rafał Miłecki <rafal@...ecki.pl>,
        "David S . Miller" <davem@...emloft.net>,
        Eric Dumazet <edumazet@...gle.com>,
        Jakub Kicinski <kuba@...nel.org>,
        Paolo Abeni <pabeni@...hat.com>,
        Frank Rowand <frowand.list@...il.com>,
        linux-mtd@...ts.infradead.org, devicetree@...r.kernel.org,
        linux-kernel@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
        netdev@...r.kernel.org, Ahmad Fatoum <a.fatoum@...gutronix.de>
Subject: Re: [PATCH v1 07/14] nvmem: core: add per-cell post processing

Hi,

Am 2022-08-30 15:37, schrieb Srinivas Kandagatla:
> On 25/08/2022 22:44, Michael Walle wrote:
>> Instead of relying on the name the consumer is using for the cell, 
>> like
>> it is done for the nvmem .cell_post_process configuration parameter,
>> provide a per-cell post processing hook. This can then be populated by
>> the NVMEM provider (or the NVMEM layout) when adding the cell.
>> 
>> Signed-off-by: Michael Walle <michael@...le.cc>
>> ---
>>   drivers/nvmem/core.c           | 16 ++++++++++++++++
>>   include/linux/nvmem-consumer.h |  5 +++++
>>   2 files changed, 21 insertions(+)
>> 
>> diff --git a/drivers/nvmem/core.c b/drivers/nvmem/core.c
>> index 5357fc378700..cbfbe6264e6c 100644
>> --- a/drivers/nvmem/core.c
>> +++ b/drivers/nvmem/core.c
>> @@ -52,6 +52,7 @@ struct nvmem_cell_entry {
>>   	int			bytes;
>>   	int			bit_offset;
>>   	int			nbits;
>> +	nvmem_cell_post_process_t post_process;
> 
> 
> two post_processing callbacks for cells is confusing tbh, we could
> totally move to use of cell->post_process.
> 
> one idea is to point cell->post_process to nvmem->cell_post_process
> during cell creation time which should clean this up a bit.

You'll then trigger the read-only check below for all the cells
if nvmem->cell_post_process is set.

> Other option is to move to using layouts for every thing.

As mentioned in a previous reply, I can't see how it could be
achieved. The problem here is that:
  (1) the layout isn't creating the cells, the OF parser is
  (2) even if we would create the cells, we wouldn't know
      which cell needs the post_process. So we are back to
      the situation above, were we have to add it to all
      the cells, making them read-only. [We depend on the
      name of the nvmem-consumer to apply the hook.

> prefixing post_process with read should also make it explicit that
> this callback is very specific to reads only.

good idea.

-michael

>>   	struct device_node	*np;
>>   	struct nvmem_device	*nvmem;
>>   	struct list_head	node;
>> @@ -468,6 +469,7 @@ static int 
>> nvmem_cell_info_to_nvmem_cell_entry_nodup(struct nvmem_device *nvmem,
>>   	cell->offset = info->offset;
>>   	cell->bytes = info->bytes;
>>   	cell->name = info->name;
>> +	cell->post_process = info->post_process;
>>     	cell->bit_offset = info->bit_offset;
>>   	cell->nbits = info->nbits;
>> @@ -1500,6 +1502,13 @@ static int __nvmem_cell_read(struct 
>> nvmem_device *nvmem,
>>   	if (cell->bit_offset || cell->nbits)
>>   		nvmem_shift_read_buffer_in_place(cell, buf);
>>   +	if (cell->post_process) {
>> +		rc = cell->post_process(nvmem->priv, id, index,
>> +					cell->offset, buf, cell->bytes);
>> +		if (rc)
>> +			return rc;
>> +	}
>> +
>>   	if (nvmem->cell_post_process) {
>>   		rc = nvmem->cell_post_process(nvmem->priv, id, index,
>>   					      cell->offset, buf, cell->bytes);
>> @@ -1608,6 +1617,13 @@ static int __nvmem_cell_entry_write(struct 
>> nvmem_cell_entry *cell, void *buf, si
>>   	    (cell->bit_offset == 0 && len != cell->bytes))
>>   		return -EINVAL;
>>   +	/*
>> +	 * Any cells which have a post_process hook are read-only because we
>> +	 * cannot reverse the operation and it might affect other cells, 
>> too.
>> +	 */
>> +	if (cell->post_process)
>> +		return -EINVAL;
> 
> Post process was always implicitly for reads only, this check should
> also tie the loose ends of cell_post_processing callback.
> 
> 
> --srini
>> +
>>   	if (cell->bit_offset || cell->nbits) {
>>   		buf = nvmem_cell_prepare_write_buffer(cell, buf, len);
>>   		if (IS_ERR(buf))
>> diff --git a/include/linux/nvmem-consumer.h 
>> b/include/linux/nvmem-consumer.h
>> index 980f9c9ac0bc..761b8ef78adc 100644
>> --- a/include/linux/nvmem-consumer.h
>> +++ b/include/linux/nvmem-consumer.h
>> @@ -19,6 +19,10 @@ struct device_node;
>>   struct nvmem_cell;
>>   struct nvmem_device;
>>   +/* duplicated from nvmem-provider.h */
>> +typedef int (*nvmem_cell_post_process_t)(void *priv, const char *id, 
>> int index,
>> +					 unsigned int offset, void *buf, size_t bytes);
>> +
>>   struct nvmem_cell_info {
>>   	const char		*name;
>>   	unsigned int		offset;
>> @@ -26,6 +30,7 @@ struct nvmem_cell_info {
>>   	unsigned int		bit_offset;
>>   	unsigned int		nbits;
>>   	struct device_node	*np;
>> +	nvmem_cell_post_process_t post_process;
>>   };
>>     /**