Date: Sat, 16 Sep 2017 18:28:46 -0400 From: Mark Johnston <markj@FreeBSD.org> To: Andreas Tobler <andreast@FreeBSD.org> Cc: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org Subject: Re: svn commit: r323290 - head/sys/vm Message-ID: <20170916222846.GA88851@bish> In-Reply-To: <67bb96f2-da01-8bce-65ba-bf811f51e56d@FreeBSD.org> References: <201709072143.v87Lhdsg060310@repo.freebsd.org> <c25a9965-b665-51ad-abe6-71beb2ae0440@FreeBSD.org> <20170914203232.GA72190@bish> <67bb96f2-da01-8bce-65ba-bf811f51e56d@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Sep 16, 2017 at 09:01:56PM +0200, Andreas Tobler wrote: > On 14.09.17 22:32, Mark Johnston wrote: > > On Thu, Sep 14, 2017 at 09:51:17PM +0200, Andreas Tobler wrote: > >> Hi Mark, > >> > >> On 07.09.17 23:43, Mark Johnston wrote: > >>> Author: markj > >>> Date: Thu Sep 7 21:43:39 2017 > >>> New Revision: 323290 > >>> URL: https://svnweb.freebsd.org/changeset/base/323290 > >>> > >>> Log: > >>> Speed up vm_page_array initialization. > >>> > >>> We currently initialize the vm_page array in three passes: one to zero > >>> the array, one to initialize the "order" field of each page (necessary > >>> when inserting them into the vm_phys buddy allocator one-by-one), and > >>> one to initialize the remaining non-zero fields and individually insert > >>> each page into the allocator. > >>> > >>> Merge the three passes into one following a suggestion from alc: > >>> initialize vm_page fields in a single pass, and use vm_phys_free_contig() > >>> to efficiently insert physical memory segments into the buddy allocator. > >>> This reduces the initialization time to a third or a quarter of what it > >>> was before on most systems that I tested. > >>> > >>> Reviewed by: alc, kib > >>> MFC after: 3 weeks > >>> Differential Revision: https://reviews.freebsd.org/D12248 > >>> > >>> Modified: > >>> head/sys/vm/vm_page.c > >>> head/sys/vm/vm_phys.c > >>> head/sys/vm/vm_phys.h > >> > >> I just found out that this commit breaks booting my powerpc64 Quad G5. > >> It hangs, pressing ctrl-t shows: cmd: sh [*vm active pagequeue]. > >> > >> Sometimes it hangs earlier when the kbd is not there yet (usb), then I > >> can't get the process/task where it hangs. > >> > >> Note, this kernel is compiled with the default gcc (4.2.1-FreeBSD) > >> > >> Any ideas how to find out what's wrong? > > > > Are you able to break into DDB when the hang occurs? If so, the output > > of "show page" would be helpful. > > Unfortunately not from the beginning. The keyboard is usb and it gets > installed late. Once it survives the loading of the kbd and co, I can > enter into ddb. But it is a trial and error. So far I didn't succeed to > come that far. > > > Are you running with INVARIANTS configured? If not, please try that. > > The above was w/o INVARIANTS. With invariants the kernel panics > immediately after boot, see pic. Thanks. Could you please try applying the patch at the end of this email and see if that fixes the issue? I have not yet tried to compile it, sorry. > > >> The previous revision, r323289 seems stable, at least it survived > >> several kernel builds. > > > > Could you apply the patch below and capture the first page or so of > > output from after the kernel starts booting? > > I applied this diff and you see its output on the pic: > > https://people.freebsd.org/~andreast/r323290_generic64_with_dbg_patch.jpg > > I try now to get that far that I have a kbd and capture a 'show page'. I don't think that's necessary anymore given the information provided in the picture. diff --git a/sys/powerpc/aim/mmu_oea.c b/sys/powerpc/aim/mmu_oea.c index 04066418fcc1..b576474fcd1d 100644 --- a/sys/powerpc/aim/mmu_oea.c +++ b/sys/powerpc/aim/mmu_oea.c @@ -287,6 +287,7 @@ boolean_t moea_is_referenced(mmu_t, vm_page_t); int moea_ts_referenced(mmu_t, vm_page_t); vm_offset_t moea_map(mmu_t, vm_offset_t *, vm_paddr_t, vm_paddr_t, int); boolean_t moea_page_exists_quick(mmu_t, pmap_t, vm_page_t); +void moea_page_init(mmu_t, vm_page_t); int moea_page_wired_mappings(mmu_t, vm_page_t); void moea_pinit(mmu_t, pmap_t); void moea_pinit0(mmu_t, pmap_t); @@ -334,6 +335,7 @@ static mmu_method_t moea_methods[] = { MMUMETHOD(mmu_ts_referenced, moea_ts_referenced), MMUMETHOD(mmu_map, moea_map), MMUMETHOD(mmu_page_exists_quick,moea_page_exists_quick), + MMUMETHOD(mmu_page_init, moea_page_init), MMUMETHOD(mmu_page_wired_mappings,moea_page_wired_mappings), MMUMETHOD(mmu_pinit, moea_pinit), MMUMETHOD(mmu_pinit0, moea_pinit0), @@ -1594,6 +1596,15 @@ moea_page_exists_quick(mmu_t mmu, pmap_t pmap, vm_page_t m) return (rv); } +void +moea_page_init(mmu_t mmu __unused, vm_page_t m) +{ + + m->md.mdpg_attrs = 0; + m->md.mdpg_cache_attrs = VM_MEMATTR_DEFAULT; + LIST_INIT(&m->md.mdpg_pvoh); +} + /* * Return the number of managed mappings to the given physical page * that are wired. diff --git a/sys/powerpc/aim/mmu_oea64.c b/sys/powerpc/aim/mmu_oea64.c index c0461ff57453..28c9c79916f1 100644 --- a/sys/powerpc/aim/mmu_oea64.c +++ b/sys/powerpc/aim/mmu_oea64.c @@ -251,6 +251,7 @@ boolean_t moea64_is_referenced(mmu_t, vm_page_t); int moea64_ts_referenced(mmu_t, vm_page_t); vm_offset_t moea64_map(mmu_t, vm_offset_t *, vm_paddr_t, vm_paddr_t, int); boolean_t moea64_page_exists_quick(mmu_t, pmap_t, vm_page_t); +void moea64_page_init(mmu_t, vm_page_t); int moea64_page_wired_mappings(mmu_t, vm_page_t); void moea64_pinit(mmu_t, pmap_t); void moea64_pinit0(mmu_t, pmap_t); @@ -298,6 +299,7 @@ static mmu_method_t moea64_methods[] = { MMUMETHOD(mmu_ts_referenced, moea64_ts_referenced), MMUMETHOD(mmu_map, moea64_map), MMUMETHOD(mmu_page_exists_quick,moea64_page_exists_quick), + MMUMETHOD(mmu_page_init, moea64_page_init), MMUMETHOD(mmu_page_wired_mappings,moea64_page_wired_mappings), MMUMETHOD(mmu_pinit, moea64_pinit), MMUMETHOD(mmu_pinit0, moea64_pinit0), @@ -1898,6 +1900,15 @@ moea64_page_exists_quick(mmu_t mmu, pmap_t pmap, vm_page_t m) return (rv); } +void +moea64_page_init(mmu_t mmu __unused, vm_page_t m) +{ + + m->md.mdpg_attrs = 0; + m->md.mdpg_cache_attrs = VM_MEMATTR_DEFAULT; + LIST_INIT(&m->md.mdpg_pvoh); +} + /* * Return the number of managed mappings to the given physical page * that are wired.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20170916222846.GA88851>