Date: Sat, 22 Aug 2015 21:15:22 +0200 From: Svatopluk Kraus <onwahe@gmail.com> To: Dmitry Marakasov <amdmi3@amdmi3.ru> Cc: Adrian Chadd <adrian.chadd@gmail.com>, "freebsd-arm@FreeBSD.org" <freebsd-arm@freebsd.org>, Ian Lepore <ian@freebsd.org> Subject: Re: Instability likely related to new pmap on Cubieboard A10 Message-ID: <CAFHCsPUMxiNHoUqxmz2TjELX943v2b8dkJzgTNowJ%2B=%2B7qDSMg@mail.gmail.com> In-Reply-To: <20150820201020.GC1245@hades.panopticon> References: <20150819013834.GD79354@hades.panopticon> <CAJ-VmokJv=nvcEkBogWeNJT65MosxnFcaitWn8FXJu32eahSdw@mail.gmail.com> <20150819120753.GH79354@hades.panopticon> <CAFHCsPVSGuWWY97ac2QVGAE77Lz2gJ12wDLpzH_kNdZsLQxh%2BQ@mail.gmail.com> <20150819134708.GJ79354@hades.panopticon> <CAFHCsPVwZS_rCnvKztg7g2%2BvrOBwQpqpPYyA2=hCLGiiU5=mrQ@mail.gmail.com> <20150819232836.GA1245@hades.panopticon> <CAFHCsPUfMhkCjiip7o6ZwGx4jNb1-Xqptsj9h_CzZ8xtfZswiA@mail.gmail.com> <20150820185417.GB1245@hades.panopticon> <CAFHCsPVrNFEDkq-j99j7a5GDBN2u7kRJRGk6Ggym2v_dre%2BN3A@mail.gmail.com> <20150820201020.GC1245@hades.panopticon>
next in thread | previous in thread | raw e-mail | index | archive | help
[-- Attachment #1 --] On Thu, Aug 20, 2015 at 10:10 PM, Dmitry Marakasov <amdmi3@amdmi3.ru> wrote: > * Svatopluk Kraus (onwahe@gmail.com) wrote: > >> >> If it helps, set it again to 1 and try attached patch, please. >> >> >> >> If neither the first nor the seconf attemp will work, can you send me >> >> output like before but from another panic. I would like to see >> >> diference. >> > >> > Just for the case, with patch and sp_enabled back to 1: >> > >> > https://people.freebsd.org/~amdmi3/pmap3.log >> > >> >> Thanks. Meantime, I tried most recent HEAD on pandaboard and >> beaglebone black and no problem there. Do you have enabled INVARIANTS >> and INVARIANT_SUPPORT in your config? > > I've enabled them at some point - at least last two runs had these > enabled. Any other way I could help? Maybe I should check if it was > new pmap commit which caused this, and if not, bisect it? > Can you try attached semi-debug patch, please? I want to be sure that problem is not on patched place. Thanks, Svata [-- Attachment #2 --] Index: sys/arm/arm/pmap-v6-new.c =================================================================== --- sys/arm/arm/pmap-v6-new.c (revision 286961) +++ sys/arm/arm/pmap-v6-new.c (working copy) @@ -1669,6 +1669,9 @@ pmap_pte2list_init(&pv_vafree, pv_chunkbase, pv_maxchunks); } +uint32_t pmap_qenter_nozero; +uint32_t pmap_qenter_valid; + /* * Add a list of wired pages to the kva * this routine is only used for temporary @@ -1681,12 +1684,10 @@ void pmap_qenter(vm_offset_t sva, vm_page_t *ma, int count) { - u_int anychanged; pt2_entry_t *epte2p, *pte2p, pte2; vm_page_t m; vm_paddr_t pa; - anychanged = 0; pte2p = pt2map_entry(sva); epte2p = pte2p + count; while (pte2p < epte2p) { @@ -1693,16 +1694,18 @@ m = *ma++; pa = VM_PAGE_TO_PHYS(m); pte2 = pte2_load(pte2p); - if ((pte2_pa(pte2) != pa) || - (pte2_attr(pte2) != m->md.pat_mode)) { - anychanged++; - pte2_store(pte2p, PTE2_KERN(pa, PTE2_AP_KRW, - m->md.pat_mode)); - } - pte2p++; + if (pte2 != 0) + pmap_qenter_nozero++; + if (pte2_is_valid(pte2)) + pmap_qenter_valid++; + if (pte2_pa(pte2) == pa && pte2_attr(pte2) != m->md.pat_mode) + panic("%s: va %#x pa %#x old %x new %x", __func__, + sva, pa, pte2_attr(pte2), m->md.pat_mode); + pte2_clear(pte2p); + tlb_flush(sva); + pte2_store(pte2p++, PTE2_KERN(pa, PTE2_AP_KRW, m->md.pat_mode)); + sva += PAGE_SIZE; } - if (__predict_false(anychanged)) - tlb_flush_range(sva, count * PAGE_SIZE); } /*
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAFHCsPUMxiNHoUqxmz2TjELX943v2b8dkJzgTNowJ%2B=%2B7qDSMg>