x86/cpu/vmware: Fixes for 5.4
Subsystems (2)
cover x86/cpu/vmware: Fixes for 5.4
From: Thomas Hellstrom <[email protected]>
Two fixes for recently introduced regressions:
Patch 1 is more or less idential to a previous patch fixing the VMW_PORT macro on LLVM's assembler. However, that patch left out the VMW_HYPERCALL macro (probably not configured for use), so let's fix that also.
Patch 2 fixes another VMW_PORT run-time regression at platform detection time
PATCH 1/2 x86/cpu/vmware: Use the full form of INL in VMWARE_HYPERCALL
From: Thomas Hellstrom <[email protected]>
LLVM's assembler doesn't accept the short form INL instruction:
inl (%%dx)
but instead insists on the output register to be explicitly specified.
This was previously fixed for the VMWARE_PORT macro. Fix it also for the VMWARE_HYPERCALL macro.
--- arch/x86/include/asm/vmware.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
arch/x86/include/asm/vmware.h +2 -1
@@ -29,7 +29,8 @@/* The low bandwidth call. The low word of edx is presumed clear. */ #define VMWARE_HYPERCALL \ - ALTERNATIVE_2("movw $" VMWARE_HYPERVISOR_PORT ", %%dx; inl (%%dx)", \ + ALTERNATIVE_2("movw $" VMWARE_HYPERVISOR_PORT \ + ", %%dx; inl (%%dx), %%eax", \ "vmcall", X86_FEATURE_VMCALL, \ "vmmcall", X86_FEATURE_VMW_VMMCALL)
Signature · 1 line
2.21.0
Re: [PATCH 1/2] x86/cpu/vmware: Use the full form of INL in VMWARE_HYPERCALL
Quoted 7 lines · Thomas Hellström (VMware)
From: Thomas Hellstrom <[email protected]>
LLVM's assembler doesn't accept the short form INL instruction:
inl (%%dx)
Quoted 8 lines
but instead insists on the output register to be explicitly specified.
This was previously fixed for the
VMWARE_PORTmacro. Fix it also for theVMWARE_HYPERCALLmacro.Fixes b4dd4f6e3648 ("Add a header file for hypercall definitions")Suggested-by Sami Tolvanen <[email protected]>Signed-off-by Thomas Hellstrom <[email protected]>
Thank you for the patch.
Quoted 31 lines
Cc "H. Peter Anvin" <[email protected]>Cc Ingo Molnar <[email protected]>Cc Thomas Gleixner <[email protected]>Cc x86-ml <[email protected]>Cc Borislav Petkov <[email protected]>--- arch/x86/include/asm/vmware.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)arch/x86/include/asm/vmware.h +2
-2@@ -29,7 +29,8 @@/* The low bandwidth call. The low word of edx is presumed clear. */ #define VMWARE_HYPERCALL \ - ALTERNATIVE_2("movw $" VMWARE_HYPERVISOR_PORT ", %%dx; inl (%%dx)", \ + ALTERNATIVE_2("movw $" VMWARE_HYPERVISOR_PORT \ + ", %%dx; inl (%%dx), %%eax", \ "vmcall", X86_FEATURE_VMCALL, \ "vmmcall", X86_FEATURE_VMW_VMMCALL) --2.21.0
--
You received this message because you are subscribed to the Google Groups "Clang Built Linux" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/clang-built-linux/20191018134052.3023-2-thomas_os%40shipmail.org.
Signature · 2 lines
Thanks, ~Nick Desaulniers
Re: [PATCH 1/2] x86/cpu/vmware: Use the full form of INL in VMWARE_HYPERCALL
In reply to Thomas Hellström (VMware)Quoted 29 lines · Thomas Hellström (VMware)
From: Thomas Hellstrom <[email protected]>
LLVM's assembler doesn't accept the short form INL instruction:
inl (%%dx)
but instead insists on the output register to be explicitly specified.
This was previously fixed for the
VMWARE_PORTmacro. Fix it also for theVMWARE_HYPERCALLmacro.Fixes b4dd4f6e3648 ("Add a header file for hypercall definitions")Suggested-by Sami Tolvanen <[email protected]>Signed-off-by Thomas Hellstrom <[email protected]>Cc "H. Peter Anvin" <[email protected]>Cc Ingo Molnar <[email protected]>Cc Thomas Gleixner <[email protected]>Cc x86-ml <[email protected]>Cc Borislav Petkov <[email protected]>--- arch/x86/include/asm/vmware.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)arch/x86/include/asm/vmware.h +0
-0
arch/x86/include/asm/vmware.h line 33
Quoted 6 lines
arch/x86/include/asm/vmware.h +2
-1@@ -29,7 +29,8 @@/* The low bandwidth call. The low word of edx is presumed clear. */ #define VMWARE_HYPERCALL \ - ALTERNATIVE_2("movw $" VMWARE_HYPERVISOR_PORT ", %%dx; inl (%%dx)", \ + ALTERNATIVE_2("movw $" VMWARE_HYPERVISOR_PORT \ + ", %%dx; inl (%%dx), %%eax", \
Why wrap in the middle of movw? Wrapping between instructions or letting the line poke out is more readable IMO, e.g.
ALTERNATIVE_2("movw $" VMWARE_HYPERVISOR_PORT ", %%dx; " \
"inl (%%dx), %%eax", \or
ALTERNATIVE_2("movw $" VMWARE_HYPERVISOR_PORT ", %%dx; inl (%%dx), %%eax", \Quoted 6 lines
arch/x86/include/asm/vmware.h +0
-0"vmcall" ,X86_FEATURE_VMCALL , \ "vmmcall", X86_FEATURE_VMW_VMMCALL)Signature · 1 line
2.21.0
PATCH 2/2 x86/cpu/vmware: Fix platform detection VMWARE_PORT macro
In reply to Thomas Hellström (VMware)From: Thomas Hellstrom <[email protected]>
The platform detection VMWARE_PORT macro uses the VMWARE_HYPERVISOR_PORT definition, but expects it to be an integer. However, when it was moved to the new vmware.h include file, it was changed to be a string to better fit into the VMWARE_HYPERCALL set of macros. This obviously breaks the platform detection VMWARE_PORT functionality.
Change the VMWARE_HYPERVISOR_PORT and VMWARE_HYPERVISOR_PORT_HB definitions to be integers, and use __stringify() for their stringified form when needed.
--- arch/x86/include/asm/vmware.h | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-)
arch/x86/include/asm/vmware.h +8 -5
@@ -4,6 +4,7 @@#include <asm/cpufeatures.h> #include <asm/alternative.h> +#include <linux/stringify.h> /* * The hypercall definitions differ in the low word of the %edx argument
@@ -20,8 +21,8 @@*/ /* Old port-based version */ -#define VMWARE_HYPERVISOR_PORT "0x5658" -#define VMWARE_HYPERVISOR_PORT_HB "0x5659" +#define VMWARE_HYPERVISOR_PORT 0x5658 +#define VMWARE_HYPERVISOR_PORT_HB 0x5659/* Current vmcall / vmmcall version */ #define VMWARE_HYPERVISOR_HB BIT(0)
@@ -29,7 +30,7 @@/* The low bandwidth call. The low word of edx is presumed clear. */ #define VMWARE_HYPERCALL \ - ALTERNATIVE_2("movw $" VMWARE_HYPERVISOR_PORT \ + ALTERNATIVE_2("movw $" __stringify(VMWARE_HYPERVISOR_PORT) \ ", %%dx; inl (%%dx), %%eax", \ "vmcall", X86_FEATURE_VMCALL, \ "vmmcall", X86_FEATURE_VMW_VMMCALL)
@@ -39,7 +40,8 @@* HB and OUT bits set. */ #define VMWARE_HYPERCALL_HB_OUT \ - ALTERNATIVE_2("movw $" VMWARE_HYPERVISOR_PORT_HB ", %%dx; rep outsb", \ + ALTERNATIVE_2("movw $" __stringify(VMWARE_HYPERVISOR_PORT_HB) \ + ", %%dx; rep outsb", \ "vmcall", X86_FEATURE_VMCALL, \ "vmmcall", X86_FEATURE_VMW_VMMCALL)
@@ -48,7 +50,8 @@* HB bit set. */ #define VMWARE_HYPERCALL_HB_IN \ - ALTERNATIVE_2("movw $" VMWARE_HYPERVISOR_PORT_HB ", %%dx; rep insb", \ + ALTERNATIVE_2("movw $" __stringify(VMWARE_HYPERVISOR_PORT_HB) \ + ", %%dx; rep insb", \ "vmcall", X86_FEATURE_VMCALL, \ "vmmcall", X86_FEATURE_VMW_VMMCALL)#endif
Signature · 1 line
2.21.0
Re: [PATCH 2/2] x86/cpu/vmware: Fix platform detection VMWARE_PORT macro
Quoted 66 lines · Thomas Hellström (VMware)
From: Thomas Hellstrom <[email protected]>
The platform detection
VMWARE_PORTmacro uses theVMWARE_HYPERVISOR_PORTdefinition, but expects it to be an integer. However, when it was moved to the newvmware.hinclude file, it was changed to be a string to better fit into theVMWARE_HYPERCALLset of macros. This obviously breaks the platform detectionVMWARE_PORTfunctionality.Change the
VMWARE_HYPERVISOR_PORTandVMWARE_HYPERVISOR_PORT_HBdefinitions to be integers, and use__stringify()for their stringified form when needed.Fixes b4dd4f6e3648 ("Add a header file for hypercall definitions")Signed-off-by Thomas Hellstrom <[email protected]>Cc "H. Peter Anvin" <[email protected]>Cc Ingo Molnar <[email protected]>Cc Thomas Gleixner <[email protected]>Cc x86-ml <[email protected]>Cc Borislav Petkov <[email protected]>--- arch/x86/include/asm/vmware.h | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-)arch/x86/include/asm/vmware.h +6
-4@@ -4,6 +4,7 @@#include <asm/cpufeatures.h> #include <asm/alternative.h> +#include <linux/stringify.h> /* * The hypercall definitions differ in the low word of the %edx argument @@ -20,8 +21,8 @@*/ /* Old port-based version */ -#define VMWARE_HYPERVISOR_PORT "0x5658" -#define VMWARE_HYPERVISOR_PORT_HB "0x5659" +#define VMWARE_HYPERVISOR_PORT 0x5658 +#define VMWARE_HYPERVISOR_PORT_HB 0x5659/* Current vmcall / vmmcall version */ #define VMWARE_HYPERVISOR_HB BIT(0)@@ -29,7 +30,7 @@/* The low bandwidth call. The low word of edx is presumed clear. */ #define VMWARE_HYPERCALL \ - ALTERNATIVE_2("movw $" VMWARE_HYPERVISOR_PORT \ + ALTERNATIVE_2("movw $" __stringify(VMWARE_HYPERVISOR_PORT) \ ", %%dx; inl (%%dx), %%eax", \ "vmcall", X86_FEATURE_VMCALL, \ "vmmcall", X86_FEATURE_VMW_VMMCALL)@@ -39,7 +40,8 @@* HB and OUT bits set. */ #define VMWARE_HYPERCALL_HB_OUT \ - ALTERNATIVE_2("movw $" VMWARE_HYPERVISOR_PORT_HB ", %%dx; rep outsb", \ + ALTERNATIVE_2("movw $" __stringify(VMWARE_HYPERVISOR_PORT_HB) \ + ", %%dx; rep outsb", \ "vmcall", X86_FEATURE_VMCALL, \ "vmmcall", X86_FEATURE_VMW_VMMCALL)
arch/x86/include/asm/vmware.h line 54
Quoted 6 lines
arch/x86/include/asm/vmware.h +2
-1@@ -48,7 +50,8 @@* HB bit set. */ #define VMWARE_HYPERCALL_HB_IN \ - ALTERNATIVE_2("movw $" VMWARE_HYPERVISOR_PORT_HB ", %%dx; rep insb", \ + ALTERNATIVE_2("movw $" __stringify(VMWARE_HYPERVISOR_PORT_HB) \ + ", %%dx; rep insb", \
Same comment on wrapping in the middle of an instruction. Wrapping after movw will stick out, but only by a char or two.
Quoted 6 lines
arch/x86/include/asm/vmware.h +0
-0"vmcall", X86_FEATURE_VMCALL, \ "vmmcall", X86_FEATURE_VMW_VMMCALL)#endif Signature · 1 line
2.21.0