On Fri, Oct 12, 2007 at 05:02:58PM +0400, Sergey Matveychuk wrote: > Kostik Belousov wrote: > >On Fri, Oct 12, 2007 at 05:20:49AM -0700, Jeremy Chadwick wrote: > >>On Fri, Oct 12, 2007 at 09:13:47AM -0300, JoaoBR wrote: > >>>somebody has a translation for this in my dmesg of amd64? > >>>SdMaP0:: AP1 6C0P.U0 0#0M1B /Lsa utnrcahnesdf!e > >>Here's the translation, and look closely: > >> > >>SMP: AP CPU #1 Launched! > >>da0: 160.000MB/s transfe > >> > >>And yes, I have seen this before on amd64. My guess is that it's two > >>things in the kernel trying to output to the console buffer at the exact > >>same time, and there's no mutex lock being done. > > > >I think that setting the option PRINTF_BUFR_SIZE 128 could help in this > >situation. > > BTW, note my kern/116310. It fixes relate issue for log() function. But > I did not touch kernel's printf(). Our logs have no junk now. I answered you somewhere already, AFAIR. I think that the patch is wrong. You cannot use the default mutex in printf(), since it is called from the arbitrary thread contexts. Also, you could get the loss of the information on the panic. PRINTF_BUFR_SIZE seems to be better solution.
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:19 UTC