Date: Fri, 11 Apr 2008 22:52:15 -0600 From: Scott Long <scottl@samsco.org> To: Julian Elischer <julian@elischer.org> Cc: FreeBSD Current <current@freebsd.org> Subject: Re: stack hogs in kernel Message-ID: <48003FFF.3040403@samsco.org> In-Reply-To: <48003E47.4090006@samsco.org> References: <48002444.4030505@elischer.org> <48003E47.4090006@samsco.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Scott Long wrote: > Julian Elischer wrote: >> >> A few years ago I took a profile of stack hogs in the kernel. >> I just redid the exercise... >> >> http://people.freebsd.org/~julian/stack-hogs.txt >> >> >> the worst offenders in MY (2 month old) kernel are: >> >> 0xc05667e3 kldstat [kernel]: 2100 >> 0xc07214f8 sendsig [kernel]: 1416 >> 0xc04fb426 ugenread [kernel]: 1200 >> 0xc070616b ipmi_smbios_identify [kernel]: 1136 >> 0xc050bd26 usbd_new_device [kernel]: 1128 >> 0xc0525a83 pfs_readlink [kernel]: 1092 >> 0xc04fb407 ugenwrite [kernel]: 1056 >> 0xc055ea33 prison_enforce_statfs [kernel]: 1044 >> 0xc0686e8a nfsrv_rename [kernel]: 984 >> 0xc05d0886 m_copymdata [kernel]: 956 >> 0xc06605a3 nfs4_mount [kernel]: 884 >> 0xc0561e25 kse_switchin [kernel]: 848 >> 0xc0680ac6 nfsrv_statfs [kernel]: 828 >> 0xc0439473 cam_periph_error [kernel]: 820 >> 0xc071ff36 sigreturn [kernel]: 812 >> 0xc067e876 nfsrv_fsinfo [kernel]: 804 >> 0xc05fe7b3 freebsd4_fhstatfs [kernel]: 804 >> 0xc0571272 sysctl_out_proc [kernel]: 804 >> 0xc067ebf7 nfsrv_fsinfo [kernel]: 804 >> 0xc0571006 sysctl_out_proc [kernel]: 804 >> 0xc044b743 cddone [kernel]: 800 >> 0xc06862cf nfsrv_link [kernel]: 796 > > I'll take a look at cddone(). I see where the problem is, but it's hard > to fix correctly. However, the call chain leading to and from cddone is > pretty small and unlikely to recurse; I seriously doubt that it will > cause problems. > Forgot to say that the stat for cam_periph_error() is due to gcc auto-inlining camperiphscsistatuserror(), which in turn auto-inlines camperiphscsisenseerror(). I wonder how many other functions on your hit-list are victimized by similar bogus gcc optimizations? Scott
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?48003FFF.3040403>