Re: [PATCH v2] powerpc/pseries: read the lpar name from the firmware

[Michael Ellerman]

Hi Laurent,

Just a few minor comments below ...

Laurent Dufour <ldufour@xxxxxxxxxxxxx> writes:
> The LPAR name may be changed after the LPAR has been started in the HMC.
> In that case lparstat command is not reporting the updated value because it
> reads it from the device tree which is read at boot time.
>
> However this value could be read from RTAS.
>
> Adding this value in the /proc/powerpc/lparcfg output allows to read the
> updated value.
>
> Cc: Nathan Lynch <nathanl@xxxxxxxxxxxxx>
> Signed-off-by: Laurent Dufour <ldufour@xxxxxxxxxxxxx>
> ---
> v2:
>  address Nathan's comments.
>  change title to partition_name aligning with existing partition_id
> ---
>  arch/powerpc/platforms/pseries/lparcfg.c | 53 ++++++++++++++++++++++++
>  1 file changed, 53 insertions(+)
>
> diff --git a/arch/powerpc/platforms/pseries/lparcfg.c b/arch/powerpc/platforms/pseries/lparcfg.c
> index f71eac74ea92..0deca7b4cd81 100644
> --- a/arch/powerpc/platforms/pseries/lparcfg.c
> +++ b/arch/powerpc/platforms/pseries/lparcfg.c
> @@ -311,6 +311,58 @@ static void parse_mpp_x_data(struct seq_file *m)
>  		seq_printf(m, "coalesce_pool_spurr=%ld\n", mpp_x_data.pool_spurr_cycles);
>  }
>  
> +/*
> + * PAPR defines no maximum for the LPAR name, and defines that the maximum
> + * length of the get-system-parameter's output buffer is 4000 plus 2 bytes for
> + * the length. Limit LPAR's name size to 1024
> + */
> +#define SPLPAR_LPAR_NAME_MAXLEN	1026

Why not just allocate the full 4000? That's not much memory in the
scheme of things, and allocating less risks us needing to update this in
future. I realise that's a ridiculously long LPAR name, but still.

> +#define SPLPAR_LPAR_NAME_TOKEN	55

Can you add a comment saying where that value is defined, PAPR?

> +static void parse_lpar_name(struct seq_file *m)

Nitpick, but we're not really parsing it, we're just reading it from firmware.

> +{
> +	int rc, len, token;
> +	unsigned char *local_buffer;
> +
> +	token = rtas_token("ibm,get-system-parameter");
> +	if (token == RTAS_UNKNOWN_SERVICE)
> +		return;
> +
> +	local_buffer = kmalloc(SPLPAR_LPAR_NAME_MAXLEN, GFP_KERNEL);
> +	if (!local_buffer)
> +		return;
> +
> +	do {
> +		spin_lock(&rtas_data_buf_lock);
> +
> +		memset(rtas_data_buf, 0, RTAS_DATA_BUF_SIZE);
> +		rc = rtas_call(token, 3, 1, NULL, SPLPAR_LPAR_NAME_TOKEN,
> +			       __pa(rtas_data_buf), RTAS_DATA_BUF_SIZE);
> +		memcpy(local_buffer, rtas_data_buf, SPLPAR_LPAR_NAME_MAXLEN);

Would be nice to skip the memcpy() if rc is non-zero.

> +		spin_unlock(&rtas_data_buf_lock);
> +	} while (rtas_busy_delay(rc));
> +
> +	if (rc != 0) {
> +		pr_err_once(
> +			"%s %s Error calling get-system-parameter (0x%x)\n",
> +			__FILE__, __func__, rc);
> +	} else {
> +		local_buffer[SPLPAR_LPAR_NAME_MAXLEN - 1] = '\0';
> +		len = local_buffer[0] * 256 + local_buffer[1];

This is kind of a hand rolled be16_to_cpu().

And we also have the + 2 to skip the length in several places.

I feel like the code would be cleaner if we had a type for the result
buffer, eg something like:

union {
	char raw_buffer[1026];
        struct {
        	__be16 len;
                char data[1024];
        };
};

> +		/*
> +		 * Forces an empty string in the weird case fw reports no data.
> +		 */
> +		if (!len)
> +			local_buffer[2] = '\0';
> +
> +		seq_printf(m, "partition_name=%s\n", local_buffer + 2);

Then above you could use the data field rather than having to add the
"2" in both places.

> +	}
> +
> +	kfree(local_buffer);
> +}
> +
> +
>  #define SPLPAR_CHARACTERISTICS_TOKEN 20
>  #define SPLPAR_MAXLENGTH 1026*(sizeof(char))
>  
> @@ -496,6 +548,7 @@ static int pseries_lparcfg_data(struct seq_file *m, void *v)
>  
>  	if (firmware_has_feature(FW_FEATURE_SPLPAR)) {
>  		/* this call handles the ibm,get-system-parameter contents */
> +		parse_lpar_name(m);
>  		parse_system_parameter_string(m);
>  		parse_ppp_data(m);
>  		parse_mpp_data(m);
> -- 
> 2.34.1


cheers