Skip to contentSkip to conversation

x86: fix brk area initialization

PATCH Superseded Juergen Gross 2 participants 8 messages Jun 22, 2022 – Jun 23, 2022

Subsystems (2)

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Juergen Gross

author maintainer
#0 lore raw

cover x86: fix brk area initialization

The brk area needs to be zeroed initially, like the .bss section.

Juergen Gross (2):
x86/xen: use clear_bss() for Xen PV guests
x86: fix setup of brk area

 arch/x86/include/asm/setup.h |  3 +++
 arch/x86/kernel/head64.c     |  4 +++-
 arch/x86/xen/enlighten_pv.c  |  8 ++++++--
 arch/x86/xen/xen-head.S      | 10 +---------
 4 files changed, 13 insertions(+), 12 deletions(-)
Signature · 1 line
2.35.3
author maintainer
#1 lore raw

PATCH 1/2 x86/xen: use clear_bss() for Xen PV guests

Instead of clearing the bss area in assembly code, use the clear_bss() function.

This requires to pass the start_info address as parameter to xen_start_kernel() in order to avoid the xen_start_info being zeroed again.

---
 arch/x86/include/asm/setup.h |  3 +++
 arch/x86/kernel/head64.c     |  2 +-
 arch/x86/xen/enlighten_pv.c  |  8 ++++++--
 arch/x86/xen/xen-head.S      | 10 +---------
 4 files changed, 11 insertions(+), 12 deletions(-)
arch/x86/include/asm/setup.h +3 -0
index f8b9ee97a891..f37cbff7354c 100644
@@ -120,6 +120,9 @@ void *extend_brk(size_t size, size_t align);
 	static char __brk_##name[size]
 
 extern void probe_roms(void);
+
+void clear_bss(void);
+
 #ifdef __i386__
 
 asmlinkage void __init i386_start_kernel(void);
arch/x86/kernel/head64.c +1 -1
index bd4a34100ed0..e7e233209a8c 100644
@@ -426,7 +426,7 @@ void __init do_early_exception(struct pt_regs *regs, int trapnr)
 
 /* Don't add a printk in there. printk relies on the PDA which is not initialized 
    yet. */
-static void __init clear_bss(void)
+void __init clear_bss(void)
 {
 	memset(__bss_start, 0,
 	       (unsigned long) __bss_stop - (unsigned long) __bss_start);
arch/x86/xen/enlighten_pv.c +6 -2
index e3297b15701c..70fb2ea85e90 100644
@@ -1183,15 +1183,19 @@ static void __init xen_domu_set_legacy_features(void)
 extern void early_xen_iret_patch(void);
 
 /* First C function to be called on Xen boot */
-asmlinkage __visible void __init xen_start_kernel(void)
+asmlinkage __visible void __init xen_start_kernel(struct start_info *si)
 {
 	struct physdev_set_iopl set_iopl;
 	unsigned long initrd_start = 0;
 	int rc;
 
-	if (!xen_start_info)
+	if (!si)
 		return;
 
+	clear_bss();
+
+	xen_start_info = si;
+
 	__text_gen_insn(&early_xen_iret_patch,
 			JMP32_INSN_OPCODE, &early_xen_iret_patch, &xen_iret,
 			JMP32_INSN_SIZE);
arch/x86/xen/xen-head.S +1 -9
index 3a2cd93bf059..13af6fe453e3 100644
@@ -48,15 +48,6 @@ SYM_CODE_START(startup_xen)
 	ANNOTATE_NOENDBR
 	cld
 
-	/* Clear .bss */
-	xor %eax,%eax
-	mov $__bss_start, %rdi
-	mov $__bss_stop, %rcx
-	sub %rdi, %rcx
-	shr $3, %rcx
-	rep stosq
-
-	mov %rsi, xen_start_info
 	mov initial_stack(%rip), %rsp
 
 	/* Set up %gs.
@@ -71,6 +62,7 @@ SYM_CODE_START(startup_xen)
 	cdq
 	wrmsr
 
+	mov	%rsi, %rdi
 	call xen_start_kernel
 SYM_CODE_END(startup_xen)
 	__FINIT
Signature · 1 line
2.35.3
Reviewed-by on v1 1/2
#2 lore raw

Re: [PATCH 1/2] x86/xen: use clear_bss() for Xen PV guests

Quoted 8 lines · Juergen Gross

Instead of clearing the bss area in assembly code, use the clear_bss() function.

This requires to pass the start_info address as parameter to xen_start_kernel() in order to avoid the xen_start_info being zeroed again.

author maintainer
#3 lore raw

PATCH 2/2 x86: fix setup of brk area

In reply to Juergen Gross

Commit e32683c6f7d2 ("x86/mm: Fix RESERVE_BRK() for older binutils") put the brk area into the .bss segment, causing it not to be cleared initially. As the brk area is used to allocate early page tables, these might contain garbage in not explicitly written entries.

This is especially a problem for Xen PV guests, as the hypervisor will validate page tables (check for writable page tables and hypervisor private bits) before accepting them to be used. There have been reports of early crashes of PV guests due to illegal page table contents.

Fix that by letting clear_bss() clear the brk area, too.

Fixes e32683c6f7d2 ("x86/mm: Fix RESERVE_BRK() for older binutils")
---
 arch/x86/kernel/head64.c | 2 ++
 1 file changed, 2 insertions(+)
arch/x86/kernel/head64.c +2 -0
index e7e233209a8c..6a3cfaf6b72a 100644
@@ -430,6 +430,8 @@ void __init clear_bss(void)
 {
 	memset(__bss_start, 0,
 	       (unsigned long) __bss_stop - (unsigned long) __bss_start);
+	memset(__brk_base, 0,
+	       (unsigned long) __brk_limit - (unsigned long) __brk_base);
 }
 
 static unsigned long get_cmd_line_ptr(void)
Signature · 1 line
2.35.3
#4 lore raw

Re: [PATCH 2/2] x86: fix setup of brk area

Quoted 3 lines · Juergen Gross

Commit e32683c6f7d2 ("x86/mm: Fix RESERVE_BRK() for older binutils") put the brk area into the .bss segment, causing it not to be cleared initially.

This reads contradictively: If the area was put in .bss, it would be cleared. Thing is it is put in .bss..brk in the object files, while the linker script puts it in .brk (i.e. outside of .bss).

Quoted 2 lines

As the brk area is used to allocate early page tables, these might contain garbage in not explicitly written entries.

I'm surprised this lack of zero-initialization didn't cause any issue outside of PV Xen. Unless of course there never was the intention for users of the facility to assume blank pages coming from there, in which case Xen's use for early page tables would have been wrong (in not explicitly zeroing the space first).

Jan

author maintainer
#5 lore raw

Re: [PATCH 2/2] x86: fix setup of brk area

Quoted 8 lines · Jan Beulich
Quoted 3 lines · Juergen Gross

Commit e32683c6f7d2 ("x86/mm: Fix RESERVE_BRK() for older binutils") put the brk area into the .bss segment, causing it not to be cleared initially.

This reads contradictively: If the area was put in .bss, it would be cleared. Thing is it is put in .bss..brk in the object files, while the linker script puts it in .brk (i.e. outside of .bss).

Hmm, yes, this should be reworded.

Quoted 9 lines
Quoted 3 lines

As the brk area is used to allocate early page tables, these might contain garbage in not explicitly written entries.

I'm surprised this lack of zero-initialization didn't cause any issue outside of PV Xen. Unless of course there never was the intention for users of the facility to assume blank pages coming from there, in which case Xen's use for early page tables would have been wrong (in not explicitly zeroing the space first).

Fun fact: Its not Xen's use for early page tables, but the kernel's init code. It is used for bare metal, too.

The use case for initial page tables is the problematic one. Only the needed page table entries are written by the kernel, so the other ones keep their initial garbage values. As normally no uninitialized entries are ever referenced, this will have no real impact.

With Xen, however, ALL entries are being validated, which led to the early crash of dom0.

Juergen

#6 lore raw

Re: [PATCH 2/2] x86: fix setup of brk area

Quoted 22 lines · Juergen Gross
Quoted 8 lines · Jan Beulich
Quoted 3 lines · Juergen Gross

Commit e32683c6f7d2 ("x86/mm: Fix RESERVE_BRK() for older binutils") put the brk area into the .bss segment, causing it not to be cleared initially.

This reads contradictively: If the area was put in .bss, it would be cleared. Thing is it is put in .bss..brk in the object files, while the linker script puts it in .brk (i.e. outside of .bss).

Hmm, yes, this should be reworded.

Quoted 9 lines
Quoted 3 lines

As the brk area is used to allocate early page tables, these might contain garbage in not explicitly written entries.

I'm surprised this lack of zero-initialization didn't cause any issue outside of PV Xen. Unless of course there never was the intention for users of the facility to assume blank pages coming from there, in which case Xen's use for early page tables would have been wrong (in not explicitly zeroing the space first).

Quoted 7 lines

Fun fact: Its not Xen's use for early page tables, but the kernel's init code. It is used for bare metal, too.

The use case for initial page tables is the problematic one. Only the needed page table entries are written by the kernel, so the other ones keep their initial garbage values. As normally no uninitialized entries are ever referenced, this will have no real impact.

Are you sure there couldn't surface user-mode accessible page table entries pointing at random pages?

Jan

author maintainer
#7 lore raw

Re: [PATCH 2/2] x86: fix setup of brk area

Quoted 26 lines · Jan Beulich
Quoted 25 lines · Juergen Gross
Quoted 8 lines · Jan Beulich
Quoted 3 lines · Juergen Gross

Commit e32683c6f7d2 ("x86/mm: Fix RESERVE_BRK() for older binutils") put the brk area into the .bss segment, causing it not to be cleared initially.

This reads contradictively: If the area was put in .bss, it would be cleared. Thing is it is put in .bss..brk in the object files, while the linker script puts it in .brk (i.e. outside of .bss).

Hmm, yes, this should be reworded.

Quoted 9 lines
Quoted 3 lines

As the brk area is used to allocate early page tables, these might contain garbage in not explicitly written entries.

I'm surprised this lack of zero-initialization didn't cause any issue outside of PV Xen. Unless of course there never was the intention for users of the facility to assume blank pages coming from there, in which case Xen's use for early page tables would have been wrong (in not explicitly zeroing the space first).

Fun fact: Its not Xen's use for early page tables, but the kernel's init code. It is used for bare metal, too.

Quoted 7 lines
Quoted 4 lines

The use case for initial page tables is the problematic one. Only the needed page table entries are written by the kernel, so the other ones keep their initial garbage values. As normally no uninitialized entries are ever referenced, this will have no real impact.

Are you sure there couldn't surface user-mode accessible page table entries pointing at random pages?

No, I'm not sure this can't happen.

Juergen