>>> On 10.12.16 at 17:43, <avscomput...@gmail.com> wrote:
> When doing "xl debug-key s; xl dmesg | tail" as instructed for checking
> TSC emulation mode on x86, I always see this line unterminated:
>
> (XEN) dom2(hvm): mode=0,ofs=0x204b1518d4d,khz=2672736,inc=1
>
> Apparently, the cause is simple: in xen/arch/x86/time.c there is
>
> static void dump_softtsc(unsigned char key)
> {
> // ... (skipped) ...
> for_each_domain ( d )
> {
> if ( is_hardware_domain(d) && d->arch.tsc_mode == TSC_MODE_DEFAULT )
> continue;
> printk("dom%u%s: mode=%d",d->domain_id,
> is_hvm_domain(d) ? "(hvm)" : "", d->arch.tsc_mode);
> if ( d->arch.vtsc_offset )
> printk(",ofs=%#"PRIx64, d->arch.vtsc_offset);
> if ( d->arch.tsc_khz )
> printk(",khz=%"PRIu32, d->arch.tsc_khz);
> if ( d->arch.incarnation )
> printk(",inc=%"PRIu32, d->arch.incarnation);
> #if !defined(NDEBUG) || defined(CONFIG_PERF_COUNTERS)
> if ( !(d->arch.vtsc_kerncount | d->arch.vtsc_usercount) )
> printk("\n");
> else
> printk(",vtsc count: %"PRIu64" kernel, %"PRIu64" user\n",
> d->arch.vtsc_kerncount, d->arch.vtsc_usercount);
> #endif
> domcnt++;
> }
> // ... (skipped) ...
> }
>
> That is, "\n" is only output in debug builds. Please, add an unconditional
> printk("\n") in #else clause for regular builds. Following is my formal
> patch proposal, but it is so trivial that you better reimplement that
> yourself as your rules mandate, with all that "Signed-off-by" and other
> cool git-related and legal-related stuff. Thanks in advance.
Please submit patches as standalone mails, i.e. without initial
problem description (other than what goes in the commit message).
Also please include [PATCH] in the subject.
> From dac74660f526436ecc6a2fa0101ec81f6822eb66 Mon Sep 17 00:00:00 2001
> From: Anton Samsonov <de...@zxlab.ru>
> Date: Sat, 10 Dec 2016 18:51:38 +0300
> Subject: [PATCH] x86/time: Fix missing "\n" in non-debug log.
>
> On x86, when dumping the emulated TSC information to dmesg, the line
>
> dom%u%s: mode=%d,ofs=0#ul,khz=%u,inc=%u
>
> is formed by several conditional statements, possibly followed by other
> printk() calls which ultimately end the line with "\n", but only in debug
> builds or with performance counters enabled, - in regular builds, however,
> the line stays unfinished and continued by further output.
>
> This patch adds such final printk("\n") for regular builds, too.
> ---
There's no Signed-off-by here, without which we won't be able to
take the patch. However, ...
> --- a/xen/arch/x86/time.c
> +++ b/xen/arch/x86/time.c
> @@ -2186,6 +2186,8 @@ static void dump_softtsc(unsigned char key)
> else
> printk(",vtsc count: %"PRIu64" kernel, %"PRIu64" user\n",
> d->arch.vtsc_kerncount, d->arch.vtsc_usercount);
> +#else
> + printk("\n");
> #endif
> domcnt++;
> }
I think this wants doing differently - I'll submit a replacement patch
in a minute, with you as the reporter of the issue.
Jan
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel
