Re: [PATCH v2 2/2] scsi: core: Eliminate scsi_log_{reserve,release}_buffer()

[Bart Van Assche <bvanassche@acm.org>]

On 9/7/26 7:44 AM, Hannes Reinecke wrote:
> On 8/31/26 7:58 PM, Bart Van Assche wrote:
>> -static size_t sdev_format_header(char *logbuf, size_t logbuf_len,
>> -                 const char *name, int tag)
>> +static void __printf(5, 0)
>> +sdev_tag_prefix_vprintk(const char *level, const struct scsi_device 
>> *sdev,
>> +            const char *name, int tag, const char *fmt,
>> +            va_list *args)
>>   {
>> -    size_t off = 0;
>> +    const struct device *dev;
>> +    struct va_format vaf = {
>> +        .fmt = fmt,
>> +        .va = args,
>> +    };
>> -    if (name)
>> -        off += scnprintf(logbuf + off, logbuf_len - off,
>> -                 "[%s] ", name);
>> +    if (!sdev)
>> +        return;
>> -    if (WARN_ON(off >= logbuf_len))
>> -        return off;
>> +    dev = &sdev->sdev_gendev;
>> +    if (name) {
>> +        if (tag >= 0)
>> +            dev_printk(level, dev, "[%s] tag#%d %pV", name, tag,
>> +                   &vaf);
>> +        else
>> +            dev_printk(level, dev, "[%s] %pV", name, &vaf);
>> +    } else {
>> +        if (tag >= 0)
>> +            dev_printk(level, dev, "tag#%d %pV", tag, &vaf);
>> +        else
>> +            dev_printk(level, dev, "%pV", &vaf);
>> +    }
>> +}
> But that would completely revert the idea of this patch; 'dev_printk' 
> will always insert a newline after the string argument, and the whole 
> point of this patch was that do _not_ want to have a newline here as we
> want to print out several things before terminating the log message.

No direct nor any indirect callers of this function use KERN_CONT, isn't
it? Please let me know if I overlooked something.

Bart.