On 15.11.2017 09:20, Paolo Bonzini wrote:
> There were two thinkos in commit 937e239 ("x86: use alloc_phys",
> 2017-11-02).
>
> First, alloc_phys is not initialized if there are no multiboot modules,
> as is the case when invoking QEMU directly with a test kernel.
>
> Second, one "&" was missing, so the topmost memory address was passed
> as the size to phys_alloc_init.
>
> Signed-off-by: Paolo Bonzini <pbonzini@xxxxxxxxxx>
> ---
> lib/x86/setup.c | 8 ++++----
> 1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/lib/x86/setup.c b/lib/x86/setup.c
> index f6a8ea8..ff7c8d3 100644
> --- a/lib/x86/setup.c
> +++ b/lib/x86/setup.c
> @@ -46,6 +46,10 @@ void setup_multiboot(struct mbi_bootinfo *bootinfo)
> {
> struct mbi_module *mods;
>
> + /* TODO: use e820 */
> + u64 end_of_memory = bootinfo->mem_upper * 1024ull;
> + phys_alloc_init((uintptr_t) &edata, end_of_memory - (uintptr_t) &edata);
> +
> if (bootinfo->mods_count != 1)
> return;
>
> @@ -53,10 +57,6 @@ void setup_multiboot(struct mbi_bootinfo *bootinfo)
>
> initrd = (char *)(uintptr_t) mods->start;
> initrd_size = mods->end - mods->start;
> -
> - /* TODO: use e820 */
> - u64 end_of_memory = bootinfo->mem_upper * 1024ull;
> - phys_alloc_init((uintptr_t) &edata, end_of_memory - (uintptr_t)edata);
> }
>
> void setup_libcflat(void)
>
Looks good to me. (unfortunately had no time yet to review the original
series)
--
Thanks,
David / dhildenb
