On Thu, Feb 13, 2025 at 07:50:42AM -0800, Dave Hansen wrote: > On 12/13/24 01:54, Yan Zhao wrote: > > + /* > > + * The destination addresses are searched from system RAM rather than > > + * being allocated from the buddy allocator, so they are not guaranteed > > + * to be accepted by the current kernel. Accept the destination > > + * addresses before kexec swaps their content with the segments' source > > + * pages to avoid accessing memory before it is accepted. > > + */ > > + for (i = 0; i < nr_segments; i++) > > + accept_memory(image->segment[i].mem, image->segment[i].memsz); > > The "searched from system RAM" phrase both here and in the changelog > doesn't quite parse for me. > > Also "System RAM" is the normal phrase that I use to describe the memory > that mostly ends up _going_ into the buddy allocator. It's not just me: > > cat /proc/iomem | grep 'System RAM' > > I think a more useful comment (and changelog) might be something like this: > > The core kernel focuses on accepting memory which is known to be > System RAM. However, there might be areas that are reserved in > the memory map, not exposed to the kernel as "System RAM" and > not accepted by firmware. Accept the memory before kexec touches > it.
If kernel compiled with CONFIG_UNACCEPTED_MEMORY, EFI_UNACCEPTED_MEMORY is part of System RAM. It translates to E820_TYPE_RAM. See setup_e820() in EFI stub. -- Kiryl Shutsemau / Kirill A. Shutemov
