Re: [PATCHv5 10/16] x86/tdx: Convert shared memory back to private on kexec

From: Nikolay Borisov
Date: Mon Jan 15 2024 - 05:53:54 EST




On 23.12.23 г. 1:52 ч., Kirill A. Shutemov wrote:
TDX guests allocate shared buffers to perform I/O. It is done by
allocating pages normally from the buddy allocator and converting them
to shared with set_memory_decrypted().

The second kernel has no idea what memory is converted this way. It only
sees E820_TYPE_RAM.

Accessing shared memory via private mapping is fatal. It leads to
unrecoverable TD exit.

On kexec walk direct mapping and convert all shared memory back to
private. It makes all RAM private again and second kernel may use it
normally.

The conversion occurs in two steps: stopping new conversions and
unsharing all memory. In the case of normal kexec, the stopping of
conversions takes place while scheduling is still functioning. This
allows for waiting until any ongoing conversions are finished. The
second step is carried out when all CPUs except one are inactive and
interrupts are disabled. This prevents any conflicts with code that may
access shared memory.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@xxxxxxxxxxxxxxx>
Reviewed-by: Rick Edgecombe <rick.p.edgecombe@xxxxxxxxx>
---
arch/x86/coco/tdx/tdx.c | 119 +++++++++++++++++++++++++++++++-
arch/x86/include/asm/x86_init.h | 2 +
arch/x86/kernel/crash.c | 6 ++
arch/x86/kernel/reboot.c | 13 ++++
4 files changed, 138 insertions(+), 2 deletions(-)

diff --git a/arch/x86/coco/tdx/tdx.c b/arch/x86/coco/tdx/tdx.c
index 8a49484a2917..5c64db168edd 100644
--- a/arch/x86/coco/tdx/tdx.c
+++ b/arch/x86/coco/tdx/tdx.c
@@ -6,8 +6,10 @@
#include <linux/cpufeature.h>
#include <linux/debugfs.h>
+#include <linux/delay.h>
#include <linux/export.h>
#include <linux/io.h>
+#include <linux/kexec.h>
#include <asm/coco.h>
#include <asm/tdx.h>
#include <asm/vmx.h>
@@ -15,6 +17,7 @@
#include <asm/insn.h>
#include <asm/insn-eval.h>
#include <asm/pgtable.h>
+#include <asm/set_memory.h>
/* MMIO direction */
#define EPT_READ 0
@@ -41,6 +44,9 @@
static atomic_long_t nr_shared;
+static atomic_t conversions_in_progress;
+static bool conversion_allowed = true;

Given the usage model of this variable, shouldn't it be simply accessed via READ/WRITE_ONCE macros?

+
static inline bool pte_decrypted(pte_t pte)
{
return cc_mkdec(pte_val(pte)) == pte_val(pte);
@@ -726,6 +732,14 @@ static bool tdx_tlb_flush_required(bool private)
static bool tdx_cache_flush_required(void)
{
+ /*
+ * Avoid issuing CLFLUSH on set_memory_decrypted() if conversions
+ * stopped. Otherwise it can race with unshare_all_memory() and trigger
+ * implicit conversion to shared.
+ */
+ if (!conversion_allowed)
+ return false;
+
/*
* AMD SME/SEV can avoid cache flushing if HW enforces cache coherence.
* TDX doesn't have such capability.
@@ -809,12 +823,25 @@ static bool tdx_enc_status_changed(unsigned long vaddr, int numpages, bool enc)
static int tdx_enc_status_change_prepare(unsigned long vaddr, int numpages,
bool enc)
{
+ atomic_inc(&conversions_in_progress);
+
+ /*
+ * Check after bumping conversions_in_progress to serialize
+ * against tdx_shutdown().
+ */
+ if (!conversion_allowed) {
+ atomic_dec(&conversions_in_progress);
+ return -EBUSY;
+ }

nit: Can you make the inc of conversions_in_progress be done here, this eliminated the atomic_dec in case they aren't. Somewhat simplifies the logic.
+
/*
* Only handle shared->private conversion here.
* See the comment in tdx_early_init().
*/
- if (enc && !tdx_enc_status_changed(vaddr, numpages, enc))
+ if (enc && !tdx_enc_status_changed(vaddr, numpages, enc)) {
+ atomic_dec(&conversions_in_progress);
return -EIO;
+ }
return 0;
}
@@ -826,17 +853,102 @@ static int tdx_enc_status_change_finish(unsigned long vaddr, int numpages,
* Only handle private->shared conversion here.
* See the comment in tdx_early_init().
*/
- if (!enc && !tdx_enc_status_changed(vaddr, numpages, enc))
+ if (!enc && !tdx_enc_status_changed(vaddr, numpages, enc)) {
+ atomic_dec(&conversions_in_progress);
return -EIO;
+ }
if (enc)
atomic_long_sub(numpages, &nr_shared);
else
atomic_long_add(numpages, &nr_shared);
+ atomic_dec(&conversions_in_progress);
+
return 0;
}
+static void tdx_kexec_stop_conversion(bool crash)
+{
+ /* Stop new private<->shared conversions */
+ conversion_allowed = false;

What's the logic behind this compiler barrier?

+ barrier();
+
+ /*
+ * Crash kernel reaches here with interrupts disabled: can't wait for
+ * conversions to finish.
+ *
+ * If race happened, just report and proceed.
+ */
+ if (!crash) {
+ unsigned long timeout;
+
+ /*
+ * Wait for in-flight conversions to complete.
+ *
+ * Do not wait more than 30 seconds.
+ */
+ timeout = 30 * USEC_PER_SEC;
+ while (atomic_read(&conversions_in_progress) && timeout--)
+ udelay(1);
+ }
+
+ if (atomic_read(&conversions_in_progress))
+ pr_warn("Failed to finish shared<->private conversions\n");
+}
+

<snip>

diff --git a/arch/x86/include/asm/x86_init.h b/arch/x86/include/asm/x86_init.h
index c9503fe2d13a..3196ff20a29e 100644
--- a/arch/x86/include/asm/x86_init.h
+++ b/arch/x86/include/asm/x86_init.h
@@ -154,6 +154,8 @@ struct x86_guest {
int (*enc_status_change_finish)(unsigned long vaddr, int npages, bool enc);
bool (*enc_tlb_flush_required)(bool enc);
bool (*enc_cache_flush_required)(void);
+ void (*enc_kexec_stop_conversion)(bool crash);
+ void (*enc_kexec_unshare_mem)(void);

These are only being initialized in the TDX case, but called in all cases when CC_ATTR_GUEST_MEM_ENCRYPT is true, which includes AMD. So it would cause a crash, no ? Shouldn't you also introduce noop handlers initialized in the default x86_platform struct in arch/x86/kernel/x86_init.c ?

};
/**
diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c
index c92d88680dbf..b99bd28ad22f 100644
--- a/arch/x86/kernel/crash.c
+++ b/arch/x86/kernel/crash.c
@@ -40,6 +40,7 @@
#include <asm/intel_pt.h>
#include <asm/crash.h>
#include <asm/cmdline.h>
+#include <asm/tdx.h>
/* Used while preparing memory map entries for second kernel */
struct crash_memmap_data {
@@ -107,6 +108,11 @@ void native_machine_crash_shutdown(struct pt_regs *regs)
crash_smp_send_stop();
+ if (cc_platform_has(CC_ATTR_GUEST_MEM_ENCRYPT)) {
+ x86_platform.guest.enc_kexec_stop_conversion(true);
+ x86_platform.guest.enc_kexec_unshare_mem();
+ }
+
cpu_emergency_disable_virtualization();
/*
diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c
index 830425e6d38e..16dde83df49a 100644
--- a/arch/x86/kernel/reboot.c
+++ b/arch/x86/kernel/reboot.c
@@ -12,6 +12,7 @@
#include <linux/delay.h>
#include <linux/objtool.h>
#include <linux/pgtable.h>
+#include <linux/kexec.h>
#include <acpi/reboot.h>
#include <asm/io.h>
#include <asm/apic.h>
@@ -31,6 +32,7 @@
#include <asm/realmode.h>
#include <asm/x86_init.h>
#include <asm/efi.h>
+#include <asm/tdx.h>
/*
* Power off function, if any
@@ -716,6 +718,14 @@ static void native_machine_emergency_restart(void)
void native_machine_shutdown(void)
{
+ /*
+ * Call enc_kexec_stop_conversion() while all CPUs are still active and
+ * interrupts are enabled. This will allow all in-flight memory
+ * conversions to finish cleanly.
+ */
+ if (cc_platform_has(CC_ATTR_GUEST_MEM_ENCRYPT) && kexec_in_progress)
+ x86_platform.guest.enc_kexec_stop_conversion(false);
+
/* Stop the cpus and apics */
#ifdef CONFIG_X86_IO_APIC
/*
@@ -752,6 +762,9 @@ void native_machine_shutdown(void)
#ifdef CONFIG_X86_64
x86_platform.iommu_shutdown();
#endif
+
+ if (cc_platform_has(CC_ATTR_GUEST_MEM_ENCRYPT) && kexec_in_progress)
+ x86_platform.guest.enc_kexec_unshare_mem();
}
static void __machine_emergency_restart(int emergency)