| Message ID | MWHPR11MB2048E2275A7815ED79E4E0628EE80@MWHPR11MB2048.namprd11.prod.outlook.com |
|---|---|
| State | New |
| Headers | show |
| Series | x86: Cache write back at 32-bit entry point | expand |
Hi Aiden, On Fri, Feb 28, 2020 at 12:54 PM Park, Aiden <aiden.park at intel.com> wrote: > > In a certain condition, invd causes cache coherence issue. > 1. Pre-stage boot code passes memory address to U-Boot > 2. The data of the memory address is still in data cache line > 3. The invd marks data cache line as invalid without write back > 4. U-Boot accesses the memory address > 5. Data is invalid > > Therefore, wbinvd is recommended at the 32-bit entry point even though > it consumes extra cpu clock cycles. > > Signed-off-by: Aiden Park <aiden.park at intel.com> > --- > arch/x86/cpu/start.S | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > This is already fixed (reverted) in the x86 tree. Would you please double check? Regards, Bin
Hi Bin, > -----Original Message----- > From: Bin Meng <bmeng.cn at gmail.com> > Sent: Thursday, February 27, 2020 9:27 PM > To: Park, Aiden <aiden.park at intel.com> > Cc: masahiroy at kernel.org; sjg at chromium.org; sr at denx.de; agraf at suse.de; u- > boot at lists.denx.de > Subject: Re: [PATCH] x86: Cache write back at 32-bit entry point > > Hi Aiden, > > On Fri, Feb 28, 2020 at 12:54 PM Park, Aiden <aiden.park at intel.com> wrote: > > > > In a certain condition, invd causes cache coherence issue. > > 1. Pre-stage boot code passes memory address to U-Boot > > 2. The data of the memory address is still in data cache line > > 3. The invd marks data cache line as invalid without write back > > 4. U-Boot accesses the memory address > > 5. Data is invalid > > > > Therefore, wbinvd is recommended at the 32-bit entry point even though > > it consumes extra cpu clock cycles. > > > > Signed-off-by: Aiden Park <aiden.park at intel.com> > > --- > > arch/x86/cpu/start.S | 2 +- > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > This is already fixed (reverted) in the x86 tree. Would you please double check? > Thank you for letting me know that. I think I missed some emails. Thanks again. Andy, appreciate your commit. I had the same issue on Slim Bootloader due to this cache coherence. > Regards, > Bin Best Regards, Aiden
diff --git a/arch/x86/cpu/start.S b/arch/x86/cpu/start.S index 26cf995db2..01524635e9 100644 --- a/arch/x86/cpu/start.S +++ b/arch/x86/cpu/start.S @@ -50,7 +50,7 @@ _x86boot_start: movl %cr0, %eax orl $(X86_CR0_NW | X86_CR0_CD), %eax movl %eax, %cr0 - invd + wbinvd /* * Zero the BIST (Built-In Self Test) value since we don't have it.
In a certain condition, invd causes cache coherence issue. 1. Pre-stage boot code passes memory address to U-Boot 2. The data of the memory address is still in data cache line 3. The invd marks data cache line as invalid without write back 4. U-Boot accesses the memory address 5. Data is invalid Therefore, wbinvd is recommended at the 32-bit entry point even though it consumes extra cpu clock cycles. Signed-off-by: Aiden Park <aiden.park at intel.com> --- arch/x86/cpu/start.S | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)