* [PATCH] x86/xenoprof: Fix kernel/user mode detection for HVM
@ 2012-10-12 16:11 Jacob Shin
2012-10-15 11:51 ` [PATCH, v2] x86/xenoprof: fix " Jan Beulich
0 siblings, 1 reply; 3+ messages in thread
From: Jacob Shin @ 2012-10-12 16:11 UTC (permalink / raw)
To: xen-devel
x86/xenoprof: Fix kernel/user mode detection for HVM
While trying oprofile under Xen, I noticed that HVM passive domain's kernel
addresses were showing up as user application. It turns out under HVM
get_cpu_user_regs()->cs contains 0x0000beef.
Signed-off-by: Jacob Shin <jacob.shin@amd.com>
diff -r e0e1350dfe9b xen/arch/x86/oprofile/xenoprof.c
--- a/xen/arch/x86/oprofile/xenoprof.c Thu Oct 11 15:57:00 2012 +0100
+++ b/xen/arch/x86/oprofile/xenoprof.c Fri Oct 12 10:48:37 2012 -0500
@@ -81,7 +81,11 @@ int xenoprofile_get_mode(const struct vc
return 2;
if ( is_hvm_vcpu(v) )
- return ((regs->cs & 3) != 3);
+ {
+ struct segment_register cs;
+ hvm_get_segment_register((struct vcpu *)v, x86_seg_cs, &cs);
+ return ((cs.sel & 3) != 3);
+ }
return guest_kernel_mode(v, regs);
}
^ permalink raw reply [flat|nested] 3+ messages in thread* [PATCH, v2] x86/xenoprof: fix kernel/user mode detection for HVM 2012-10-12 16:11 [PATCH] x86/xenoprof: Fix kernel/user mode detection for HVM Jacob Shin @ 2012-10-15 11:51 ` Jan Beulich 2012-10-15 12:09 ` Keir Fraser 0 siblings, 1 reply; 3+ messages in thread From: Jan Beulich @ 2012-10-15 11:51 UTC (permalink / raw) To: xen-devel; +Cc: Jacob Shin [-- Attachment #1: Type: text/plain, Size: 1720 bytes --] While trying oprofile under Xen, I noticed that HVM passive domain's kernel addresses were showing up as user application. It turns out under HVM get_cpu_user_regs()->cs contains 0x0000beef. Signed-off-by: Jacob Shin <jacob.shin@amd.com> Don't cast away const-ness. Use SS instead of CS to determine ring. Special-case real and protected mode. Signed-off-by: Jan Beulich <jbeulich@suse.com> --- a/xen/arch/x86/oprofile/xenoprof.c +++ b/xen/arch/x86/oprofile/xenoprof.c @@ -74,16 +74,26 @@ int compat_oprof_arch_counter(XEN_GUEST_ return 0; } -int xenoprofile_get_mode(const struct vcpu *v, - const struct cpu_user_regs *regs) +int xenoprofile_get_mode(struct vcpu *curr, const struct cpu_user_regs *regs) { if ( !guest_mode(regs) ) return 2; - if ( is_hvm_vcpu(v) ) - return ((regs->cs & 3) != 3); + if ( !is_hvm_vcpu(curr) ) + return guest_kernel_mode(curr, regs); - return guest_kernel_mode(v, regs); + switch ( hvm_guest_x86_mode(curr) ) + { + struct segment_register ss; + + case 0: /* real mode */ + return 1; + case 1: /* vm86 mode */ + return 0; + default: + hvm_get_segment_register(curr, x86_seg_ss, &ss); + return (ss.sel & 3) != 3; + } } /* --- a/xen/include/asm-x86/xenoprof.h +++ b/xen/include/asm-x86/xenoprof.h @@ -51,7 +51,7 @@ struct cpu_user_regs; void ibs_init(void); extern u32 ibs_caps; -int xenoprofile_get_mode(const struct vcpu *, const struct cpu_user_regs *); +int xenoprofile_get_mode(struct vcpu *, const struct cpu_user_regs *); static inline int xenoprof_backtrace_supported(void) { [-- Attachment #2: x86-oprof-hvm-mode.patch --] [-- Type: text/plain, Size: 1770 bytes --] x86/xenoprof: fix kernel/user mode detection for HVM While trying oprofile under Xen, I noticed that HVM passive domain's kernel addresses were showing up as user application. It turns out under HVM get_cpu_user_regs()->cs contains 0x0000beef. Signed-off-by: Jacob Shin <jacob.shin@amd.com> Don't cast away const-ness. Use SS instead of CS to determine ring. Special-case real and protected mode. Signed-off-by: Jan Beulich <jbeulich@suse.com> --- a/xen/arch/x86/oprofile/xenoprof.c +++ b/xen/arch/x86/oprofile/xenoprof.c @@ -74,16 +74,26 @@ int compat_oprof_arch_counter(XEN_GUEST_ return 0; } -int xenoprofile_get_mode(const struct vcpu *v, - const struct cpu_user_regs *regs) +int xenoprofile_get_mode(struct vcpu *curr, const struct cpu_user_regs *regs) { if ( !guest_mode(regs) ) return 2; - if ( is_hvm_vcpu(v) ) - return ((regs->cs & 3) != 3); + if ( !is_hvm_vcpu(curr) ) + return guest_kernel_mode(curr, regs); - return guest_kernel_mode(v, regs); + switch ( hvm_guest_x86_mode(curr) ) + { + struct segment_register ss; + + case 0: /* real mode */ + return 1; + case 1: /* vm86 mode */ + return 0; + default: + hvm_get_segment_register(curr, x86_seg_ss, &ss); + return (ss.sel & 3) != 3; + } } /* --- a/xen/include/asm-x86/xenoprof.h +++ b/xen/include/asm-x86/xenoprof.h @@ -51,7 +51,7 @@ struct cpu_user_regs; void ibs_init(void); extern u32 ibs_caps; -int xenoprofile_get_mode(const struct vcpu *, const struct cpu_user_regs *); +int xenoprofile_get_mode(struct vcpu *, const struct cpu_user_regs *); static inline int xenoprof_backtrace_supported(void) { [-- Attachment #3: Type: text/plain, Size: 126 bytes --] _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel ^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH, v2] x86/xenoprof: fix kernel/user mode detection for HVM 2012-10-15 11:51 ` [PATCH, v2] x86/xenoprof: fix " Jan Beulich @ 2012-10-15 12:09 ` Keir Fraser 0 siblings, 0 replies; 3+ messages in thread From: Keir Fraser @ 2012-10-15 12:09 UTC (permalink / raw) To: Jan Beulich, xen-devel; +Cc: Jacob Shin On 15/10/2012 12:51, "Jan Beulich" <JBeulich@suse.com> wrote: > While trying oprofile under Xen, I noticed that HVM passive domain's > kernel addresses were showing up as user application. It turns out > under HVM get_cpu_user_regs()->cs contains 0x0000beef. > > Signed-off-by: Jacob Shin <jacob.shin@amd.com> > > Don't cast away const-ness. Use SS instead of CS to determine ring. > Special-case real and protected mode. > > Signed-off-by: Jan Beulich <jbeulich@suse.com> Acked-by: Keir Fraser <keir@xen.org> > --- a/xen/arch/x86/oprofile/xenoprof.c > +++ b/xen/arch/x86/oprofile/xenoprof.c > @@ -74,16 +74,26 @@ int compat_oprof_arch_counter(XEN_GUEST_ > return 0; > } > > -int xenoprofile_get_mode(const struct vcpu *v, > - const struct cpu_user_regs *regs) > +int xenoprofile_get_mode(struct vcpu *curr, const struct cpu_user_regs *regs) > { > if ( !guest_mode(regs) ) > return 2; > > - if ( is_hvm_vcpu(v) ) > - return ((regs->cs & 3) != 3); > + if ( !is_hvm_vcpu(curr) ) > + return guest_kernel_mode(curr, regs); > > - return guest_kernel_mode(v, regs); > + switch ( hvm_guest_x86_mode(curr) ) > + { > + struct segment_register ss; > + > + case 0: /* real mode */ > + return 1; > + case 1: /* vm86 mode */ > + return 0; > + default: > + hvm_get_segment_register(curr, x86_seg_ss, &ss); > + return (ss.sel & 3) != 3; > + } > } > > /* > --- a/xen/include/asm-x86/xenoprof.h > +++ b/xen/include/asm-x86/xenoprof.h > @@ -51,7 +51,7 @@ struct cpu_user_regs; > void ibs_init(void); > extern u32 ibs_caps; > > -int xenoprofile_get_mode(const struct vcpu *, const struct cpu_user_regs *); > +int xenoprofile_get_mode(struct vcpu *, const struct cpu_user_regs *); > > static inline int xenoprof_backtrace_supported(void) > { > > > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xen.org > http://lists.xen.org/xen-devel ^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2012-10-15 12:09 UTC | newest] Thread overview: 3+ messages (download: mbox.gz follow: Atom feed -- links below jump to the message on this page -- 2012-10-12 16:11 [PATCH] x86/xenoprof: Fix kernel/user mode detection for HVM Jacob Shin 2012-10-15 11:51 ` [PATCH, v2] x86/xenoprof: fix " Jan Beulich 2012-10-15 12:09 ` Keir Fraser
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).