Date: Tue, 29 Jul 2014 13:08:46 +0000 (UTC) From: Marius Strobl <marius@FreeBSD.org> To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-stable@freebsd.org, svn-src-stable-9@freebsd.org Subject: svn commit: r269236 - stable/9/sys/i386/i386 Message-ID: <201407291308.s6TD8kL5066969@svn.freebsd.org>
next in thread | raw e-mail | index | archive | help
Author: marius Date: Tue Jul 29 13:08:46 2014 New Revision: 269236 URL: http://svnweb.freebsd.org/changeset/base/269236 Log: MFC: r269050 - Copying and zeroing pages via temporary mappings involves updating the corresponding page tables followed by accesses to the pages in question. This sequence is subject to the situation exactly described in the "AMD64 Architecture Programmer's Manual Volume 2: System Programming" rev. 3.23, "7.3.1 Special Coherency Considerations" [1, p. 171 f.]. Therefore, issuing the INVLPG right after modifying the PTE bits is crucial. For pmap_copy_page(), this has been broken in r124956 and later on carried over to pmap_copy_pages() derived from the former, while all other places in the i386 PMAP code use the correct order of instructions in this regard. Fixing the latter breakage solves the problem of data corruption seen with unmapped I/O enabled when running at least bare metal on AMD R-268D APUs. However, this might also fix similar corruption reported for virtualized environments. - In pmap_copy_pages(), correctly set the cache bits on the source page being copied. This change is thought to be a NOP for the real world, though. [2] 1: http://amd-dev.wpengine.netdna-cdn.com/wordpress/media/2012/10/24593_APM_v21.pdf Submitted by: kib [2] Reviewed by: alc, kib Sponsored by: Bally Wulff Games & Entertainment GmbH Modified: stable/9/sys/i386/i386/pmap.c Directory Properties: stable/9/sys/ (props changed) Modified: stable/9/sys/i386/i386/pmap.c ============================================================================== --- stable/9/sys/i386/i386/pmap.c Tue Jul 29 13:08:37 2014 (r269235) +++ stable/9/sys/i386/i386/pmap.c Tue Jul 29 13:08:46 2014 (r269236) @@ -1302,6 +1302,13 @@ pmap_pte_release(pt_entry_t *pte) mtx_unlock(&PMAP2mutex); } +/* + * NB: The sequence of updating a page table followed by accesses to the + * corresponding pages is subject to the situation described in the "AMD64 + * Architecture Programmer's Manual Volume 2: System Programming" rev. 3.23, + * "7.3.1 Special Coherency Considerations". Therefore, issuing the INVLPG + * right after modifying the PTE bits is crucial. + */ static __inline void invlcaddr(void *caddr) { @@ -4249,12 +4256,12 @@ pmap_copy_page(vm_page_t src, vm_page_t if (*sysmaps->CMAP2) panic("pmap_copy_page: CMAP2 busy"); sched_pin(); - invlpg((u_int)sysmaps->CADDR1); - invlpg((u_int)sysmaps->CADDR2); *sysmaps->CMAP1 = PG_V | VM_PAGE_TO_PHYS(src) | PG_A | pmap_cache_bits(src->md.pat_mode, 0); + invlcaddr(sysmaps->CADDR1); *sysmaps->CMAP2 = PG_V | PG_RW | VM_PAGE_TO_PHYS(dst) | PG_A | PG_M | pmap_cache_bits(dst->md.pat_mode, 0); + invlcaddr(sysmaps->CADDR2); bcopy(sysmaps->CADDR1, sysmaps->CADDR2, PAGE_SIZE); *sysmaps->CMAP1 = 0; *sysmaps->CMAP2 = 0; @@ -4282,8 +4289,6 @@ pmap_copy_pages(vm_page_t ma[], vm_offse panic("pmap_copy_pages: CMAP2 busy"); sched_pin(); while (xfersize > 0) { - invlpg((u_int)sysmaps->CADDR1); - invlpg((u_int)sysmaps->CADDR2); a_pg = ma[a_offset >> PAGE_SHIFT]; a_pg_offset = a_offset & PAGE_MASK; cnt = min(xfersize, PAGE_SIZE - a_pg_offset); @@ -4291,9 +4296,11 @@ pmap_copy_pages(vm_page_t ma[], vm_offse b_pg_offset = b_offset & PAGE_MASK; cnt = min(cnt, PAGE_SIZE - b_pg_offset); *sysmaps->CMAP1 = PG_V | VM_PAGE_TO_PHYS(a_pg) | PG_A | - pmap_cache_bits(b_pg->md.pat_mode, 0); + pmap_cache_bits(a_pg->md.pat_mode, 0); + invlcaddr(sysmaps->CADDR1); *sysmaps->CMAP2 = PG_V | PG_RW | VM_PAGE_TO_PHYS(b_pg) | PG_A | PG_M | pmap_cache_bits(b_pg->md.pat_mode, 0); + invlcaddr(sysmaps->CADDR2); a_cp = sysmaps->CADDR1 + a_pg_offset; b_cp = sysmaps->CADDR2 + b_pg_offset; bcopy(a_cp, b_cp, cnt);
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201407291308.s6TD8kL5066969>