Skip to contentSkip to conversation

x86/tdx: Support vmalloc() for tdx_enc_status_changed()

PATCH In review Dexuan Cui 2 participants 4 messages May 21, 2024 – Jun 28, 2024

Subsystems (5)
lore
Share & tools
lore

Dexuan Cui

author maintainer
#0 lore raw

PATCH 1/1 x86/tdx: Support vmalloc() for tdx_enc_status_changed()

When a TDX guest runs on Hyper-V, the hv_netvsc driver's netvsc_init_buf() allocates buffers using vzalloc(), and needs to share the buffers with the host OS by calling set_memory_decrypted(), which is not working for vmalloc() yet. Add the support by handling the pages one by one.

---

This is basically a repost of the second patch of the 2023 patchset:
https://lwn.net/ml/linux-kernel/[email protected]/

The first patch of the patchset got merged into mainline, but unluckily the second patch didn't, and I kind of lost track of it. Sorry.

Changes since the previous patchset (please refer to the link above):
Added Rick's and Dave's Reviewed-by.
Added Kai's Acked-by.
Removeda the test "if (offset_in_page(start) != 0)" since we know the 'start' is page-aligned: see __set_memory_enc_pgtable().

Please review. Thanks!
Dexuan

 arch/x86/coco/tdx/tdx.c | 35 ++++++++++++++++++++++++++++-------
 1 file changed, 28 insertions(+), 7 deletions(-)
arch/x86/coco/tdx/tdx.c +28 -7
index c1cb90369915b..abf3cd591afd3 100644
@@ -7,6 +7,7 @@
 #include <linux/cpufeature.h>
 #include <linux/export.h>
 #include <linux/io.h>
+#include <linux/mm.h>
 #include <asm/coco.h>
 #include <asm/tdx.h>
 #include <asm/vmx.h>
@@ -778,6 +779,19 @@ static bool tdx_map_gpa(phys_addr_t start, phys_addr_t end, bool enc)
 	return false;
 }
 
+static bool tdx_enc_status_changed_phys(phys_addr_t start, phys_addr_t end,
+					bool enc)
+{
+	if (!tdx_map_gpa(start, end, enc))
+		return false;
+
+	/* shared->private conversion requires memory to be accepted before use */
+	if (enc)
+		return tdx_accept_memory(start, end);
+
+	return true;
+}
+
 /*
  * Inform the VMM of the guest's intent for this physical page: shared with
  * the VMM or private to the guest.  The VMM is expected to change its mapping
@@ -785,15 +799,22 @@ static bool tdx_map_gpa(phys_addr_t start, phys_addr_t end, bool enc)
  */
 static bool tdx_enc_status_changed(unsigned long vaddr, int numpages, bool enc)
 {
-	phys_addr_t start = __pa(vaddr);
-	phys_addr_t end   = __pa(vaddr + numpages * PAGE_SIZE);
+	unsigned long start = vaddr;
+	unsigned long end = start + numpages * PAGE_SIZE;
+	unsigned long step = end - start;
+	unsigned long addr;
 
-	if (!tdx_map_gpa(start, end, enc))
-		return false;
+	/* Step through page-by-page for vmalloc() mappings */
+	if (is_vmalloc_addr((void *)vaddr))
+		step = PAGE_SIZE;
 
-	/* shared->private conversion requires memory to be accepted before use */
-	if (enc)
-		return tdx_accept_memory(start, end);
+	for (addr = start; addr < end; addr += step) {
+		phys_addr_t start_pa = slow_virt_to_phys((void *)addr);
+		phys_addr_t end_pa   = start_pa + step;
+
+		if (!tdx_enc_status_changed_phys(start_pa, end_pa, enc))
+			return false;
+	}
 
 	return true;
 }
Signature · 1 line
2.25.1
author maintainer
#1 lore raw
Quoted 27 lines

From: Dexuan Cui <[email protected]>
Sent: Monday, May 20, 2024 7:13 PM
[....]
When a TDX guest runs on Hyper-V, the hv_netvsc driver's netvsc_init_buf()
allocates buffers using vzalloc(), and needs to share the buffers with the host OS by calling set_memory_decrypted(), which is not working for vmalloc() yet. Add the support by handling the pages one by one.

Reviewed-by Kuppuswamy Sathyanarayanan

<[email protected]>

---

This is basically a repost of the second patch of the 2023 patchset:
https://lwn.net/ml/linux-kernel/[email protected]/

The first patch of the patchset got merged into mainline, but unluckily the second patch didn't, and I kind of lost track of it. Sorry.

Quoted 8 lines

Changes since the previous patchset (please refer to the link above):
Added Rick's and Dave's Reviewed-by.
Added Kai's Acked-by.
Removeda the test "if (offset_in_page(start) != 0)" since we know the 'start' is page-aligned: see __set_memory_enc_pgtable().

Please review. Thanks!
Dexuan

The patch still applies cleanly to 6.10-rc4.

A gentle ping.

#2 lore raw
In reply to Dexuan Cui
Quoted 20 lines · Dexuan Cui
(quoted diff) +7 -7
@@ -785,15 +799,22 @@ static bool tdx_map_gpa(phys_addr_t start, phys_addr_t end, bool enc)
  */
 static bool tdx_enc_status_changed(unsigned long vaddr, int numpages, bool enc)
 {
-	phys_addr_t start = __pa(vaddr);
-	phys_addr_t end   = __pa(vaddr + numpages * PAGE_SIZE);
+	unsigned long start = vaddr;
+	unsigned long end = start + numpages * PAGE_SIZE;
+	unsigned long step = end - start;
+	unsigned long addr;
 
-	if (!tdx_map_gpa(start, end, enc))
-		return false;
+	/* Step through page-by-page for vmalloc() mappings */
+	if (is_vmalloc_addr((void *)vaddr))
+		step = PAGE_SIZE;
 
-	/* shared->private conversion requires memory to be accepted before use */
-	if (enc)
-		return tdx_accept_memory(start, end);

Review static bool tdx_map_gpa(phys_addr_t start, phys_addr_t end, bool enc) · line 820

Quoted 10 lines
(quoted diff) +7 -0
+	for (addr = start; addr < end; addr += step) {
+		phys_addr_t start_pa = slow_virt_to_phys((void *)addr);
+		phys_addr_t end_pa   = start_pa + step;
+
+		if (!tdx_enc_status_changed_phys(start_pa, end_pa, enc))
+			return false;
+	}
 
 	return true;
 }

This patch collied with kexec changes. tdx_kexec_finish() calls tdx_enc_status_changed() after clearing pte, so slow_virt_to_phys() crashes on in.

Daxuan, could you check if the fixup below works for you on vmalloc addresses?

arch/x86/coco/tdx/tdx.c +9 -2
index ef8ec2425998..5e455c883bcc 100644
@@ -813,8 +813,15 @@ static bool tdx_enc_status_changed(unsigned long vaddr, int numpages, bool enc)
 		step = PAGE_SIZE;
 
 	for (addr = start; addr < end; addr += step) {
-		phys_addr_t start_pa = slow_virt_to_phys((void *)addr);
-		phys_addr_t end_pa   = start_pa + step;
+		phys_addr_t start_pa;
+		phys_addr_t end_pa;
+
+		if (virt_addr_valid(addr))
+			start_pa = __pa(addr);
+		else
+			start_pa = slow_virt_to_phys((void *)addr);
+
+		end_pa = start_pa + step;
 
 		if (!tdx_enc_status_changed_phys(start_pa, end_pa, enc))
 			return false;
Signature · 1 line
  Kiryl Shutsemau / Kirill A. Shutemov
author maintainer
#3 lore raw
Quoted 35 lines

From: Kirill A. Shutemov <[email protected]>
Sent: Friday, June 28, 2024 3:05 AM
To: Dexuan Cui <[email protected]>

Quoted 2 lines

[...]
static bool tdx_enc_status_changed(unsigned long vaddr, int numpages,

[...]
This patch collied with kexec changes. tdx_kexec_finish() calls tdx_enc_status_changed() after clearing pte, so slow_virt_to_phys() crashes on in.

Daxuan, could you check if the fixup below works for you on vmalloc addresses?

arch/x86/coco/tdx/tdx.c +9 -2
index ef8ec2425998..5e455c883bcc 100644
@@ -813,8 +813,15 @@ static bool tdx_enc_status_changed(unsigned
 long vaddr, int numpages, bool enc)
 		step = PAGE_SIZE;
 
 	for (addr = start; addr < end; addr += step) {
-		phys_addr_t start_pa = slow_virt_to_phys((void *)addr);
-		phys_addr_t end_pa   = start_pa + step;
+		phys_addr_t start_pa;
+		phys_addr_t end_pa;
+
+		if (virt_addr_valid(addr))
+			start_pa = __pa(addr);
+		else
+			start_pa = slow_virt_to_phys((void *)addr);
+
+		end_pa = start_pa + step;
 
 		if (!tdx_enc_status_changed_phys(start_pa, end_pa, enc))
Quoted 3 lines

return false;
--
Kiryl Shutsemau / Kirill A. Shutemov

Hi Kirill, your fixup works for me.

BTW, I just realized that virt_addr_valid() returns false for a vmalloc'd address.

Thanks,
Dexuan