linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
From: Dave Hansen <dave@linux.vnet.ibm.com>
To: linux-kernel@vger.kernel.org
Cc: linux-mm@kvack.org, Gleb Natapov <gleb@redhat.com>,
	Avi Kivity <avi@redhat.com>,
	Dave Hansen <dave@linux.vnet.ibm.com>
Subject: [RFCv2][PATCH 3/3] make DEBUG_VIRTUAL work earlier in boot
Date: Fri, 07 Dec 2012 16:30:25 -0500	[thread overview]
Message-ID: <20121207213025.53E4BD1B@kernel.stglabs.ibm.com> (raw)
In-Reply-To: <20121207213023.AA3AFF11@kernel.stglabs.ibm.com>


The KVM code has some repeated bugs in it around use of __pa() on
per-cpu data.  Those data are not in an area on which __pa() is
valid.  However, they are also called early enough in boot that
__vmalloc_start_set is not set, and thus the CONFIG_DEBUG_VIRTUAL
debugging does not catch them.

This adds a check to also verify them against max_low_pfn, which
we can use earler in boot than is_vmalloc_addr().  However, if
we are super-early in boot, max_low_pfn=0 and this will trip
on every call, so also make sure that max_low_pfn is set.

With this patch applied, CONFIG_DEBUG_VIRTUAL will actually
catch the bug I was chasing.

I'd love to find a generic way so that any __pa() call on percpu
areas could do a BUG_ON(), but there don't appear to be any nice
and easy ways to check if an address is a percpu one.  Anybody
have ideas on a way to do this?

---

 linux-2.6.git-dave/arch/x86/mm/numa.c     |    2 +-
 linux-2.6.git-dave/arch/x86/mm/pat.c      |    4 ++--
 linux-2.6.git-dave/arch/x86/mm/physaddr.c |   10 ++++++++--
 3 files changed, 11 insertions(+), 5 deletions(-)

diff -puN arch/x86/mm/physaddr.c~make-DEBUG_VIRTUAL-work-earlier-in-boot arch/x86/mm/physaddr.c
--- linux-2.6.git/arch/x86/mm/physaddr.c~make-DEBUG_VIRTUAL-work-earlier-in-boot	2012-11-30 16:18:44.522847232 -0500
+++ linux-2.6.git-dave/arch/x86/mm/physaddr.c	2012-11-30 16:18:44.530847298 -0500
@@ -1,3 +1,4 @@
+#include <linux/bootmem.h>
 #include <linux/mmdebug.h>
 #include <linux/module.h>
 #include <linux/mm.h>
@@ -41,16 +42,21 @@ bool __virt_addr_valid(unsigned long x)
 	return pfn_valid(x >> PAGE_SHIFT);
 }
 EXPORT_SYMBOL(__virt_addr_valid);
-
 #else
 
 #ifdef CONFIG_DEBUG_VIRTUAL
 unsigned long __phys_addr(unsigned long x)
 {
+	unsigned long phys_addr = x - PAGE_OFFSET;
 	/* VMALLOC_* aren't constants  */
 	VIRTUAL_BUG_ON(x < PAGE_OFFSET);
 	VIRTUAL_BUG_ON(__vmalloc_start_set && is_vmalloc_addr((void *) x));
-	return x - PAGE_OFFSET;
+	/* max_low_pfn is set early, but not _that_ early */
+	if (max_low_pfn) {
+		VIRTUAL_BUG_ON((phys_addr >> PAGE_SHIFT) > max_low_pfn);
+		BUG_ON(slow_virt_to_phys((void *)x) != phys_addr);
+	}
+	return phys_addr;
 }
 EXPORT_SYMBOL(__phys_addr);
 #endif
diff -puN arch/x86/kernel/kvmclock.c~make-DEBUG_VIRTUAL-work-earlier-in-boot arch/x86/kernel/kvmclock.c
diff -L sr -puN /dev/null /dev/null
diff -puN arch/x86/include/asm/page_32.h~make-DEBUG_VIRTUAL-work-earlier-in-boot arch/x86/include/asm/page_32.h
diff -puN arch/x86/mm/numa.c~make-DEBUG_VIRTUAL-work-earlier-in-boot arch/x86/mm/numa.c
--- linux-2.6.git/arch/x86/mm/numa.c~make-DEBUG_VIRTUAL-work-earlier-in-boot	2012-11-30 16:18:44.526847265 -0500
+++ linux-2.6.git-dave/arch/x86/mm/numa.c	2012-11-30 16:18:44.534847331 -0500
@@ -219,7 +219,7 @@ static void __init setup_node_data(int n
 	 */
 	nd = alloc_remap(nid, nd_size);
 	if (nd) {
-		nd_pa = __pa(nd);
+		nd_pa = __phys_addr_nodebug(nd);
 		remapped = true;
 	} else {
 		nd_pa = memblock_alloc_nid(nd_size, SMP_CACHE_BYTES, nid);
diff -puN arch/x86/mm/pat.c~make-DEBUG_VIRTUAL-work-earlier-in-boot arch/x86/mm/pat.c
--- linux-2.6.git/arch/x86/mm/pat.c~make-DEBUG_VIRTUAL-work-earlier-in-boot	2012-11-30 16:19:34.371258739 -0500
+++ linux-2.6.git-dave/arch/x86/mm/pat.c	2012-11-30 16:22:38.528778740 -0500
@@ -560,10 +560,10 @@ int kernel_map_sync_memtype(u64 base, un
 {
 	unsigned long id_sz;
 
-	if (base >= __pa(high_memory))
+	if (base > __pa(high_memory-1))
 		return 0;
 
-	id_sz = (__pa(high_memory) < base + size) ?
+	id_sz = (__pa(high_memory-1) <= base + size) ?
 				__pa(high_memory) - base :
 				size;
 
_

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

  parent reply	other threads:[~2012-12-07 21:30 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-12-07 21:30 [RFCv2][PATCH 1/3] create slow_virt_to_phys() Dave Hansen
2012-12-07 21:30 ` [RFCv2][PATCH 2/3] fix kvm's use of __pa() on percpu areas Dave Hansen
2012-12-07 21:30 ` Dave Hansen [this message]
2012-12-09 14:06 ` [RFCv2][PATCH 1/3] create slow_virt_to_phys() Gleb Natapov

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20121207213025.53E4BD1B@kernel.stglabs.ibm.com \
    --to=dave@linux.vnet.ibm.com \
    --cc=avi@redhat.com \
    --cc=gleb@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).