On Tue, Mar 05, 2019 at 12:35:48 +0100, Ján Tomko wrote:
On Tue, Mar 05, 2019 at 10:40:31AM +0100, Jiri Denemark wrote:
The log message may be useful when debugging why a specific CPU model was selected for a given set of CPUID data.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com> ---
Notes: Version 2: - separated from 11/26 cpu_x86: Allow multiple signatures for a CPU model - signature formatting code moved into a dedicated function
src/cpu/cpu_x86.c | 27 +++++++++++++++++++++++++++ 1 file changed, 27 insertions(+)
diff --git a/src/cpu/cpu_x86.c b/src/cpu/cpu_x86.c index 08677ef7ff..5a1071de4d 100644 --- a/src/cpu/cpu_x86.c +++ b/src/cpu/cpu_x86.c @@ -1773,6 +1773,26 @@ x86ModelHasSignature(virCPUx86ModelPtr model, }
+static char * +x86FormatSignatures(virCPUx86ModelPtr model) +{ + virBuffer buf = VIR_BUFFER_INITIALIZER; + size_t i; + + for (i = 0; i < model->nsignatures; i++) { + virBufferAsprintf(&buf, "%06lx,", + (unsigned long)model->signatures[i]); + } + + virBufferTrim(&buf, ",", -1); + + if (virBufferCheckError(&buf) < 0) + return NULL; + + return virBufferContentAndReset(&buf); +} + + /* * Checks whether a candidate model is a better fit for the CPU data than the * current model. @@ -1896,6 +1916,7 @@ x86Decode(virCPUDefPtr cpu, virCPUx86Data features = VIR_CPU_X86_DATA_INIT; virCPUx86VendorPtr vendor; virDomainCapsCPUModelPtr hvModel = NULL; + char *sigs = NULL;
VIR_AUTOFREE(char *)
I fixed this and pushed this series. Thanks for the reviews. Jirka