Hi,
Can someone review this patch? It's a very simple one. Applied on top of commit 9b30973d7db7572dc8bae301a1f8259992e6d215 using `git am`
I'm Leah from the Libreboot project
On 16/05/2021 00:25, Leah Rowe via SeaBIOS wrote:
Hi,
Can someone review this patch? It's a very simple one. Applied on top of commit 9b30973d7db7572dc8bae301a1f8259992e6d215 using `git am`
I'm Leah from the Libreboot project
Sorry, I forgot to add -s when doing commit.
Newly attached patch has signed-off-by line
On Sun, May 16, 2021 at 12:25:27AM +0100, Leah Rowe via SeaBIOS wrote:
From anecdotal reports, sometimes it is useful; I've heard of instances where certain Nvidia graphics cards can, if using the nouveau driver on a GNU+Linux system, be used without running any option ROM, but where the option ROM has certain data tables in it that the driver can use.
If set to 1, this will disable *all* option ROM loading, even SeaVGABIOS. It also affects CSM. In practise, the only time someone would ever use this is if they were using a graphics card in the above scenario. Such a use-case is incredibly rare.
If set to 0, the normal behaviour can be observed (load, and execute). The run time option etc/pci-optionrom-exec will still be in effect.
Unfortunately it's not valid, in the general case, to deploy an option rom and not run it. The option rom itself can alter its size field and roms are known to do that. Without running the rom, it's often not possible to get its size.
If it's not desirable to actually run the rom, you might want to consider hex editing the particular rom binary so that it's entry point just immediately returns.
-Kevin
On 20/05/2021 19:15, Kevin O'Connor wrote:
Unfortunately it's not valid, in the general case, to deploy an option rom and not run it. The option rom itself can alter its size field and roms are known to do that. Without running the rom, it's often not possible to get its size.
If it's not desirable to actually run the rom, you might want to consider hex editing the particular rom binary so that it's entry point just immediately returns.
-Kevin
Thanks. Yes, I did wonder about that, thank you for confirming it.
I haven't even tested the patch yet, but I was curious what someone would think of it.