Re: [PATCH] PCI/sysfs: add write attribute for boot_vga

[李真能]

在 2021/9/27 上午4:00, Krzysztof Wilczyński 写道:
> [+cc Huacai and Kai-Heng as they are working in this area]
>
> Hi,
>
> Thank you for sending the patch over.
>
> I assume this is simply a resend (rather than a v2), as I see no code
> changes from the previous version you sent some time ago.
Sorry, I haven't receive any reply about this email, so I resend this.
> > Add writing attribute for boot_vga sys node,
> > so we can config default video display
> > output dynamically when there are two video
> > cards on a machine.
> That's OK, but why are you adding this?  What problem does it solve and
> what is the intended user here?  Might be worth adding a little bit more
> details about why this new sysfs attribute is needed.
Xorg will detemine which graphics is prime output device according 
boot_vga, if there are two graphics card, and we want xorg output 
display to diffent graphics card, we can echo 1 to boot_vga.
> I also need to ask, as I am not sure myself, whether this is OK to do after
> booting during runtime?  What do you think Bjorn, Huacai and Kai-Heng?

I have test this function, during runtime, if xorg's graphics output 
device is card A, then we echo 1 to boot_vga of card B, and then restart 
xorg, xorg will output to card B, if we want xorg always output to card 
B, we can echo 1 to boot_vga of card B before xorg started in daemon 
process.

> Also, please correctly capitalise the subject - have a look at previous
> commit messages to see how it should look like.
>
> > +static ssize_t boot_vga_store(struct device *dev, struct device_attribute *attr,
> > +			      const char *buf, size_t count)
> > +{
> > +	unsigned long val;
> > +	struct pci_dev *pdev = to_pci_dev(dev);
> > +	struct pci_dev *vga_dev = vga_default_device();
> > +
> > +	if (kstrtoul(buf, 0, &val) < 0)
> > +		return -EINVAL;
> > +
> > +	if (val != 1)
> > +		return -EINVAL;
> Since this is a completely new API have a look at kstrtobool() over
> kstrtoul() as the former was created to handle user input more
> consistently.
As I want restrict available value  to 1, if we use  kstrtobool(), it 
will be available when user input other value.
> > +	if (!capable(CAP_SYS_ADMIN))
> > +		return -EPERM;
> > +
> Check for CAP_SYS_ADMIN is a good idea, but it has to take place before you
> attempt to accept and parse a input from the user.
>
> 	Krzysztof