Skip to contentSkip to conversation

x86-64: improve e820_search_gap()

PATCH In review Jan Beulich 3 participants 4 messages Mar 12, 2009 – Mar 13, 2009 x86

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 x86-64: improve e820_search_gap()

Blindly putting the gap close after max_pfn might be fine for native (though even there I'm uncertain regarding memory hotplug), but will certainly present a problem on Xen. And properly searching for a gap above 4Gb shouldn't hurt native.

Also, make the function static to ensure there are no other users that could depend on the previous behavior regarding the way start_addr gets specified.

Cc Jeremy Fitzhardinge <[email protected]>
---
 arch/x86/include/asm/e820.h |    2 --
 arch/x86/kernel/e820.c      |   11 +++++++++--
 2 files changed, 9 insertions(+), 4 deletions(-)
arch/x86/include/asm/e820.h +0 -2
@@ -79,8 +79,6 @@ extern u64 e820_remove_range(u64 start, 
 			     int checktype);
 extern void update_e820(void);
 extern void e820_setup_gap(void);
-extern int e820_search_gap(unsigned long *gapstart, unsigned long *gapsize,
-			unsigned long start_addr, unsigned long long end_addr);
 struct setup_data;
 extern void parse_e820_ext(struct setup_data *data, unsigned long pa_data);
 
arch/x86/kernel/e820.c +9 -2
@@ -533,13 +533,19 @@ static void __init update_e820_saved(voi
 /*
  * Search for a gap in the e820 memory space from start_addr to end_addr.
  */
-__init int e820_search_gap(unsigned long *gapstart, unsigned long *gapsize,
+static int __init
+e820_search_gap(unsigned long *gapstart, unsigned long *gapsize,
 		unsigned long start_addr, unsigned long long end_addr)
 {
 	unsigned long long last;
 	int i = e820.nr_map;
 	int found = 0;
 
+#ifdef CONFIG_X86_64
+	if (start_addr >= MAX_GAP_END)
+		last = end_addr ?: (1UL << boot_cpu_data.x86_phys_bits);
+	else
+#endif
 	last = (end_addr && end_addr < MAX_GAP_END) ? end_addr : MAX_GAP_END;
 
 	while (--i >= 0) {
@@ -585,11 +591,12 @@ __init void e820_setup_gap(void)
 
 #ifdef CONFIG_X86_64
 	if (!found) {
-		gapstart = (max_pfn << PAGE_SHIFT) + 1024*1024;
 		printk(KERN_ERR "PCI: Warning: Cannot find a gap in the 32bit "
 		       "address range\n"
 		       KERN_ERR "PCI: Unassigned devices with 32bit resource "
 		       "registers may break!\n");
+		found = e820_search_gap(&gapstart, &gapsize, MAX_GAP_END, 0);
+		BUG_ON(!found);
 	}
 #endif
 
#1 lore raw
Quoted 39 lines · Jan Beulich

Blindly putting the gap close after max_pfn might be fine for native (though even there I'm uncertain regarding memory hotplug), but will certainly present a problem on Xen. And properly searching for a gap above 4Gb shouldn't hurt native.

Also, make the function static to ensure there are no other users that could depend on the previous behavior regarding the way start_addr gets specified.

Cc Jeremy Fitzhardinge <[email protected]>
---
 arch/x86/include/asm/e820.h |    2 --
 arch/x86/kernel/e820.c      |   11 +++++++++--
 2 files changed, 9 insertions(+), 4 deletions(-)
arch/x86/include/asm/e820.h +0 -2
@@ -79,8 +79,6 @@ extern u64 e820_remove_range(u64 start, 
 			     int checktype);
 extern void update_e820(void);
 extern void e820_setup_gap(void);
-extern int e820_search_gap(unsigned long *gapstart, unsigned long *gapsize,
-			unsigned long start_addr, unsigned long long end_addr);
 struct setup_data;
 extern void parse_e820_ext(struct setup_data *data, unsigned long pa_data);
 
arch/x86/kernel/e820.c +2 -1
@@ -533,13 +533,19 @@ static void __init update_e820_saved(voi
 /*
  * Search for a gap in the e820 memory space from start_addr to end_addr.
  */
-__init int e820_search_gap(unsigned long *gapstart, unsigned long *gapsize,
+static int __init
+e820_search_gap(unsigned long *gapstart, unsigned long *gapsize,
 		unsigned long start_addr, unsigned long long end_addr)
 {

arch/x86/kernel/e820.c static void __init update_e820_saved(voi · line 549

Quoted 10 lines
arch/x86/kernel/e820.c +5 -0
 	unsigned long long last;
 	int i = e820.nr_map;
 	int found = 0;
 
+#ifdef CONFIG_X86_64
+	if (start_addr >= MAX_GAP_END)
+		last = end_addr ?: (1UL << boot_cpu_data.x86_phys_bits);
+	else
+#endif
 	last = (end_addr && end_addr < MAX_GAP_END) ? end_addr : MAX_GAP_END;

hm, this #ifdef block looks quite ugly and should be cleaned up. x86_phys_bits could be filled in on 32-bit too - and on 32-bit start_addr cannot be larger than 4GB anyway.

arch/x86/kernel/e820.c __init void e820_setup_gap(void) · line 599

Quoted 12 lines
arch/x86/kernel/e820.c +2 -1
 	while (--i >= 0) {
@@ -585,11 +591,12 @@ __init void e820_setup_gap(void)
 
 #ifdef CONFIG_X86_64
 	if (!found) {
-		gapstart = (max_pfn << PAGE_SHIFT) + 1024*1024;
 		printk(KERN_ERR "PCI: Warning: Cannot find a gap in the 32bit "
 		       "address range\n"
 		       KERN_ERR "PCI: Unassigned devices with 32bit resource "
 		       "registers may break!\n");
+		found = e820_search_gap(&gapstart, &gapsize, MAX_GAP_END, 0);
+		BUG_ON(!found);

that BUG_ON() will be hard to debug - please use a WARN_ON instead.

Ingo

#2 lore raw
Quoted 11 lines
Quoted 1 line
Quoted 1 line

Ingo Molnar <[email protected]> 12.03.09 12:02 >>>

arch/x86/kernel/e820.c +5 -0
+#ifdef CONFIG_X86_64
+	if (start_addr >= MAX_GAP_END)
+		last = end_addr ?: (1UL << boot_cpu_data.x86_phys_bits);
+	else
+#endif
 	last = (end_addr && end_addr < MAX_GAP_END) ? end_addr : MAX_GAP_END;

hm, this #ifdef block looks quite ugly and should be cleaned up. x86_phys_bits could be filled in on 32-bit too - and on 32-bit

I'm about to submit a patch to that effect. But I'm trying to keep patches independent as much as possible.

Quoted 1 line

start_addr cannot be larger than 4GB anyway.

Correct, but I think the code would be less self-documenting if it relied on that fact rather than making clear from the first glance that the conditional is only being evaluated (and hence can only be true) on 64-bits.

Quoted 14 lines

arch/x86/kernel/e820.c __init void e820_setup_gap(void) · line 599

Quoted 11 lines
arch/x86/kernel/e820.c +2 -1
@@ -585,11 +591,12 @@ __init void e820_setup_gap(void)
 
 #ifdef CONFIG_X86_64
 	if (!found) {
-		gapstart = (max_pfn << PAGE_SHIFT) + 1024*1024;
 		printk(KERN_ERR "PCI: Warning: Cannot find a gap in the 32bit "
 		       "address range\n"
 		       KERN_ERR "PCI: Unassigned devices with 32bit resource "
 		       "registers may break!\n");
+		found = e820_search_gap(&gapstart, &gapsize, MAX_GAP_END, 0);
+		BUG_ON(!found);

that BUG_ON() will be hard to debug - please use a WARN_ON instead.

Will do, but please clarify the above point before I re-submit.

Jan

#3 lore raw
In reply to Jan Beulich
Quoted 5 lines · Jan Beulich

Blindly putting the gap close after max_pfn might be fine for native (though even there I'm uncertain regarding memory hotplug), but will certainly present a problem on Xen. And properly searching for a gap above 4Gb shouldn't hurt native.

What Xen-related issue are you thinking of?

When booting dom0 I reashape the guest memory to match the host E820 map, and for domU my plan is to leave a hole at 3-4GB for pci passthrough resources.

J