| Message ID | 8082FF9BCB2B054996454E47167FF4EC1C59CA0C@SHSMSX104.ccr.corp.intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
>>> On 10.08.17 at 13:09, <xiong.y.zhang@intel.com> wrote: > On my SKL/KBL machine, upstream 4.12 kernel dom0 couldn't boot up using > xen.efi which is xen 4.9 > > > (1) Upstream 4.11 kernel doesn't have such issue. > > (2) Upstream 4.12 kernel on my native uefi machine could boot up. > > After some debug, I have some finding: > firmware/efi.c: Reinit efi global variable > > efi is a global variable, some of efi members is initialized to > INVALID_TABLE_ADDR when efi is defined. But efi is zero when > setup_arch() begin running for unknown reason. Well, I don't think that's a precise description, but anyway the problem has been discussed before, and a patch is available (but may not have landed in the 4.12.x stable tree yet, as originally it was assumed to only be cosmetic) - look for "x86/xen/efi: Initialize only the EFI struct members used by Xen". Jan
On Thu, Aug 10, 2017 at 11:09:20AM +0000, Zhang, Xiong Y wrote: > On my SKL/KBL machine, upstream 4.12 kernel dom0 couldn't boot up using xen.efi which is xen 4.9 > > > (1) Upstream 4.11 kernel doesn't have such issue. > > (2) Upstream 4.12 kernel on my native uefi machine could boot up. > > After some debug, I have some finding: > firmware/efi.c: Reinit efi global variable > > efi is a global variable, some of efi members is initialized to > INVALID_TABLE_ADDR when efi is defined. But efi is zero when > setup_arch() begin running for unknown reason. > > This patch set it again to avoid kernel panic. This fix 4.12 kernel dom0 > couldn't boot up on xen.efi 4.9. > > diff --git a/drivers/firmware/efi/efi.c b/drivers/firmware/efi/efi.c > index b372aad..27bbcb4 100644 > --- a/drivers/firmware/efi/efi.c > +++ b/drivers/firmware/efi/efi.c > @@ -474,6 +474,9 @@ int __init efi_config_parse_tables(void *config_tables, int count, int sz, > void *tablep; > int i; > > + efi.mps = efi.acpi = efi.acpi20 = efi.smbios = efi.smbios3 = efi.sal_systab = > + efi.boot_info = efi.hcdp = efi.uga = efi.uv_systab = efi.fw_vendor = > + efi.esrt= efi.properties_table = efi.rng_seed = efi.mem_attr_table = EFI_INVALID_TABLE_ADDR; > > why efi is zero at setup_arch() begin running ? who will clear it ? See arch/x86/xen/efi.c > > thanks > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xen.org > https://lists.xen.org/xen-devel
diff --git a/drivers/firmware/efi/efi.c b/drivers/firmware/efi/efi.c index b372aad..27bbcb4 100644 --- a/drivers/firmware/efi/efi.c +++ b/drivers/firmware/efi/efi.c @@ -474,6 +474,9 @@ int __init efi_config_parse_tables(void *config_tables, int count, int sz, void *tablep; int i; + efi.mps = efi.acpi = efi.acpi20 = efi.smbios = efi.smbios3 = efi.sal_systab = + efi.boot_info = efi.hcdp = efi.uga = efi.uv_systab = efi.fw_vendor = + efi.esrt= efi.properties_table = efi.rng_seed = efi.mem_attr_table = EFI_INVALID_TABLE_ADDR; why efi is zero at setup_arch() begin running ? who will clear it ?
On my SKL/KBL machine, upstream 4.12 kernel dom0 couldn't boot up using xen.efi which is xen 4.9 (1) Upstream 4.11 kernel doesn't have such issue. (2) Upstream 4.12 kernel on my native uefi machine could boot up. After some debug, I have some finding: firmware/efi.c: Reinit efi global variable efi is a global variable, some of efi members is initialized to INVALID_TABLE_ADDR when efi is defined. But efi is zero when setup_arch() begin running for unknown reason. This patch set it again to avoid kernel panic. This fix 4.12 kernel dom0 couldn't boot up on xen.efi 4.9. thanks