5.8.7-rc1 review

This is the start of the stable review cycle for the 5.8.7 release. There are 17 patches in this series, all will be posted as a response to this one. If anyone has any issues with these being applied, please let me know.

Responses should be made by Sun, 06 Sep 2020 12:02:48 +0000. Anything received after that time might be too late.

The whole patch series can be found in one patch at:

	https://www.kernel.org/pub/linux/kernel/v5.x/stable-review/patch-5.8.7-rc1.gz

or in the git tree and branch at:
git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable-rc.git linux-5.8.y
and the diffstat can be found below.

thanks,

greg k-h

-------------
Pseudo-Shortlog of commits:

Greg Kroah-Hartman <[email protected]>

    Linux 5.8.7-rc1

Bodo Stroesser <[email protected]>
scsi: target: tcmu: Optimize use of flush_dcache_page

Johannes Berg <[email protected]>

    nl80211: fix NL80211_ATTR_HE_6GHZ_CAPABILITY usage

Sowjanya Komatineni <[email protected]>
sdhci: tegra: Remove SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK for Tegra186

Sowjanya Komatineni <[email protected]>
sdhci: tegra: Remove SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK for Tegra210

Sowjanya Komatineni <[email protected]>
arm64: tegra: Add missing timeout clock to Tegra210 SDMMC

Sowjanya Komatineni <[email protected]>
arm64: tegra: Add missing timeout clock to Tegra186 SDMMC nodes

Sowjanya Komatineni <[email protected]>
arm64: tegra: Add missing timeout clock to Tegra194 SDMMC nodes

Sowjanya Komatineni <[email protected]>
dt-bindings: mmc: tegra: Add tmclk for Tegra210 and later

James Morse <[email protected]>
KVM: arm64: Survive synchronous exceptions caused by AT instructions

James Morse <[email protected]>
KVM: arm64: Add kvm_extable for vaxorcism code

Peilin Ye <[email protected]>
media: media/v4l2-core: Fix kernel-infoleak in video_put_user()

Kim Phillips <[email protected]>
perf record/stat: Explicitly call out event modifiers in the documentation

Andy Lutomirski <[email protected]>
selftests/x86/test_vsyscall: Improve the process_vm_readv() test

Dave Hansen <[email protected]>
mm: fix pin vs. gup mismatch with gate pages

Stefano Brivio <[email protected]>
netfilter: nft_set_rbtree: Handle outcomes of tree rotations in overlap detection

Marc Zyngier <[email protected]>
HID: core: Sanitize event code and type when mapping input

Marc Zyngier <[email protected]>
HID: core: Correctly handle ReportSize being zero

-------------

Diffstat:

 .../bindings/mmc/nvidia,tegra20-sdhci.txt          | 32 ++++++++++-
 Makefile                                           |  4 +-
 arch/arm64/boot/dts/nvidia/tegra186.dtsi           | 20 ++++---
 arch/arm64/boot/dts/nvidia/tegra194.dtsi           | 15 +++--
 arch/arm64/boot/dts/nvidia/tegra210.dtsi           | 20 ++++---
 arch/arm64/include/asm/kvm_asm.h                   | 43 ++++++++++++++
 arch/arm64/kernel/vmlinux.lds.S                    |  8 +++
 arch/arm64/kvm/hyp/entry.S                         | 15 +++--
 arch/arm64/kvm/hyp/hyp-entry.S                     | 65 ++++++++++++++--------
 arch/arm64/kvm/hyp/switch.c                        | 39 +++++++++++--
 drivers/hid/hid-core.c                             | 15 ++++-
 drivers/hid/hid-input.c                            |  4 ++
 drivers/hid/hid-multitouch.c                       |  2 +
 drivers/media/v4l2-core/v4l2-ioctl.c               | 50 +++++++++--------
 drivers/mmc/host/sdhci-tegra.c                     |  2 -
 drivers/target/target_core_user.c                  | 11 ++--
 include/linux/hid.h                                | 42 +++++++++-----
 mm/gup.c                                           |  2 +-
 net/netfilter/nft_set_rbtree.c                     | 23 +++++---
 net/wireless/nl80211.c                             |  2 +-
 tools/perf/Documentation/perf-record.txt           |  4 ++
 tools/perf/Documentation/perf-stat.txt             |  4 ++
 tools/testing/selftests/x86/test_vsyscall.c        | 22 +++++++-
 23 files changed, 326 insertions(+), 118 deletions(-)

From: Marc Zyngier <[email protected]>

commit bce1305c0ece3dc549663605e567655dd701752c upstream.

It appears that a ReportSize value of zero is legal, even if a bit non-sensical. Most of the HID code seems to handle that gracefully, except when computing the total size in bytes. When fed as input to memset, this leads to some funky outcomes.

Detect the corner case and correctly compute the size.

---
 drivers/hid/hid-core.c |   15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

@@ -1598,6 +1598,17 @@ static void hid_output_field(const struc
 }
 
 /*
+ * Compute the size of a report.
+ */
+static size_t hid_compute_report_size(struct hid_report *report)
+{
+	if (report->size)
+		return ((report->size - 1) >> 3) + 1;
+
+	return 0;
+}
+
+/*
  * Create a report. 'data' has to be allocated using
  * hid_alloc_report_buf() so that it has proper size.
  */

@@ -1609,7 +1620,7 @@ void hid_output_report(struct hid_report
 	if (report->id > 0)
 		*data++ = report->id;
 
-	memset(data, 0, ((report->size - 1) >> 3) + 1);
+	memset(data, 0, hid_compute_report_size(report));
 	for (n = 0; n < report->maxfield; n++)
 		hid_output_field(report->device, report->field[n], data);
 }

@@ -1739,7 +1750,7 @@ int hid_report_raw_event(struct hid_devi
 		csize--;
 	}
 
-	rsize = ((report->size - 1) >> 3) + 1;
+	rsize = hid_compute_report_size(report);
 
 	if (report_enum->numbered && rsize >= HID_MAX_BUFFER_SIZE)
 		rsize = HID_MAX_BUFFER_SIZE - 1;

From: Marc Zyngier <[email protected]>

commit 35556bed836f8dc07ac55f69c8d17dce3e7f0e25 upstream.

When calling into hid_map_usage(), the passed event code is blindly stored as is, even if it doesn't fit in the associated bitmap.

This event code can come from a variety of sources, including devices masquerading as input devices, only a bit more "programmable".

Instead of taking the event code at face value, check that it actually fits the corresponding bitmap, and if it doesn't:
- spit out a warning so that we know which device is acting up
- NULLify the bitmap pointer so that we catch unexpected uses

Code paths that can make use of untrusted inputs can now check that the mapping was indeed correct and bail out if not.

---
 drivers/hid/hid-input.c      |    4 ++++
 drivers/hid/hid-multitouch.c |    2 ++
 include/linux/hid.h          |   42 +++++++++++++++++++++++++++++-------------
 3 files changed, 35 insertions(+), 13 deletions(-)

@@ -1132,6 +1132,10 @@ static void hidinput_configure_usage(str
 	}
 
 mapped:
+	/* Mapping failed, bail out */
+	if (!bit)
+		return;
+
 	if (device->driver->input_mapped &&
 	    device->driver->input_mapped(device, hidinput, field, usage,
 					 &bit, &max) < 0) {

@@ -856,6 +856,8 @@ static int mt_touch_input_mapping(struct
 			code = BTN_0  + ((usage->hid - 1) & HID_USAGE);
 
 		hid_map_usage(hi, usage, bit, max, EV_KEY, code);
+		if (!*bit)
+			return -1;
 		input_set_capability(hi->input, EV_KEY, code);
 		return 1;
 

From: Stefano Brivio <[email protected]>

commit 226a88de473e475cb9f993682a1c7d0c2b451ad8 upstream.

Checks for partial overlaps on insertion assume that end elements are always descendant nodes of their corresponding start, because they are inserted later. However, this is not the case if a previous delete operation caused a tree rotation as part of rebalancing.

Taking the issue reported by Andreas Fischer as an example, if we omit delete operations, the existing procedure works because, equivalently, we are inserting a start item with value 40 in the this region of the red-black tree with single-sized intervals:

                                  overlap flag
                   10 (start)
                  /  \            false
                      20 (start)
                     /  \         false
                         30 (start)
                        /  \      false
                            60 (start)
                           /  \   false
                         50 (end)
                        /  \      false
                      20 (end)
                     /  \         false
                         40 (start)

if we now delete interval 30 - 30, the tree can be rearranged in a way similar to this (note the rotation involving 50 - 50):

                                  overlap flag
                   10 (start)
                  /  \            false
                      20 (start)
                     /  \         false
                         25 (start)
                        /  \      false
                            70 (start)
                           /  \   false
                         50 (end)
                        /  \      true (from rule a1.)
                      50 (start)
                     /  \         true
                   40 (start)

and we traverse interval 50 - 50 from the opposite direction compared to what was expected.

To deal with those cases, add a start-before-start rule, b4., that covers traversal of existing intervals from the right.

We now need to restrict start-after-end rule b3. to cases where there are no occurring nodes between existing start and end elements, because addition of rule b4. isn't sufficient to ensure that the pre-existing end element we encounter while descending the tree corresponds to a start element of an interval that we already traversed entirely.

Different types of overlap detection on trees with rotations resulting from re-balancing will be covered by nft test case sets/0044interval_overlap_1.

Bugzilla: https://bugzilla.netfilter.org/show_bug.cgi?id=1449

---
 net/netfilter/nft_set_rbtree.c |   23 ++++++++++++++---------
 1 file changed, 14 insertions(+), 9 deletions(-)

@@ -238,21 +238,27 @@ static int __nft_rbtree_insert(const str
 	 *
 	 * b1. _ _ __>|  !_ _ __|  (insert end before existing start)
 	 * b2. _ _ ___|  !_ _ _>|  (insert end after existing start)
-	 * b3. _ _ ___! >|_ _ __|  (insert start after existing end)
+	 * b3. _ _ ___! >|_ _ __|  (insert start after existing end, as a leaf)
+	 *            '--' no nodes falling in this range
+	 * b4.          >|_ _   !  (insert start before existing start)
 	 *
 	 * Case a3. resolves to b3.:
 	 * - if the inserted start element is the leftmost, because the '0'
 	 *   element in the tree serves as end element
-	 * - otherwise, if an existing end is found. Note that end elements are
-	 *   always inserted after corresponding start elements.
+	 * - otherwise, if an existing end is found immediately to the left. If
+	 *   there are existing nodes in between, we need to further descend the
+	 *   tree before we can conclude the new start isn't causing an overlap
+	 *
+	 * or to b4., which, preceded by a3., means we already traversed one or
+	 * more existing intervals entirely, from the right.
 	 *
 	 * For a new, rightmost pair of elements, we'll hit cases b3. and b2.,
 	 * in that order.
 	 *
 	 * The flag is also cleared in two special cases:
 	 *
-	 * b4. |__ _ _!|<_ _ _   (insert start right before existing end)
-	 * b5. |__ _ >|!__ _ _   (insert end right after existing start)
+	 * b5. |__ _ _!|<_ _ _   (insert start right before existing end)
+	 * b6. |__ _ >|!__ _ _   (insert end right after existing start)
 	 *
 	 * which always happen as last step and imply that no further
 	 * overlapping is possible.

@@ -272,7 +278,7 @@ static int __nft_rbtree_insert(const str
 			if (nft_rbtree_interval_start(new)) {
 				if (nft_rbtree_interval_end(rbe) &&
 				    nft_set_elem_active(&rbe->ext, genmask) &&
-				    !nft_set_elem_expired(&rbe->ext))
+				    !nft_set_elem_expired(&rbe->ext) && !*p)
 					overlap = false;
 			} else {
 				overlap = nft_rbtree_interval_end(rbe) &&

@@ -288,10 +294,9 @@ static int __nft_rbtree_insert(const str
 					  nft_set_elem_active(&rbe->ext,
 							      genmask) &&
 					  !nft_set_elem_expired(&rbe->ext);
-			} else if (nft_rbtree_interval_end(rbe) &&
-				   nft_set_elem_active(&rbe->ext, genmask) &&
+			} else if (nft_set_elem_active(&rbe->ext, genmask) &&
 				   !nft_set_elem_expired(&rbe->ext)) {
-				overlap = true;
+				overlap = nft_rbtree_interval_end(rbe);
 			}
 		} else {
 			if (nft_rbtree_interval_end(rbe) &&

From: Dave Hansen <[email protected]>

commit 9fa2dd946743ae6f30dc4830da19147bf100a7f2 upstream.

Gate pages were missed when converting from get to pin_user_pages(). This can lead to refcount imbalances. This is reliably and quickly reproducible running the x86 selftests when vsyscall=emulate is enabled (the default). Fix by using try_grab_page() with appropriate flags passed.

The long story:

Today, pin_user_pages() and get_user_pages() are similar interfaces for manipulating page reference counts. However, "pins" use a "bias" value and manipulate the actual reference count by 1024 instead of 1 used by plain "gets".

That means that pin_user_pages() must be matched with unpin_user_pages() and can't be mixed with a plain put_user_pages() or put_page().

Enter gate pages, like the vsyscall page. They are pages usually in the kernel image, but which are mapped to userspace. Userspace is allowed access to them, including interfaces using get/pin_user_pages(). The refcount of these kernel pages is manipulated just like a normal user page on the get/pin side so that the put/unpin side can work the same for normal user pages or gate pages.

get_gate_page() uses try_get_page() which only bumps the refcount by 1, not 1024, even if called in the pin_user_pages() path. If someone pins a gate page, this happens:

	pin_user_pages()
		get_gate_page()
			try_get_page() // bump refcount +1
	... some time later
	unpin_user_pages()
		page_ref_sub_and_test(page, 1024))

... and boom, we get a refcount off by 1023. This is reliably and quickly reproducible running the x86 selftests when booted with vsyscall=emulate (the default). The selftests use ptrace(), but I suspect anything using pin_user_pages() on gate pages could hit this.

To fix it, simply use try_grab_page() instead of try_get_page(), and pass 'gup_flags' in so that FOLL_PIN can be respected.

This bug traces back to the very beginning of the FOLL_PIN support in commit 3faa52c03f44 ("mm/gup: track FOLL_PIN pages"), which showed up in the 5.7 release.

---
 mm/gup.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

@@ -843,7 +843,7 @@ static int get_gate_page(struct mm_struc
 			goto unmap;
 		*page = pte_page(*pte);
 	}
-	if (unlikely(!try_get_page(*page))) {
+	if (unlikely(!try_grab_page(*page, gup_flags))) {
 		ret = -ENOMEM;
 		goto unmap;
 	}

From: Andy Lutomirski <[email protected]>

commit 8891adc61dce2a8a41fc0c23262b681c3ec4b73a upstream.

The existing code accepted process_vm_readv() success or failure as long as it didn't return garbage. This is too weak: if the vsyscall page is readable, then process_vm_readv() should succeed and, if the page is not readable, then it should fail.

---
 tools/testing/selftests/x86/test_vsyscall.c |   22 ++++++++++++++++++++--
 1 file changed, 20 insertions(+), 2 deletions(-)

@@ -462,6 +462,17 @@ static int test_vsys_x(void)
 	return 0;
 }
 
+/*
+ * Debuggers expect ptrace() to be able to peek at the vsyscall page.
+ * Use process_vm_readv() as a proxy for ptrace() to test this.  We
+ * want it to work in the vsyscall=emulate case and to fail in the
+ * vsyscall=xonly case.
+ *
+ * It's worth noting that this ABI is a bit nutty.  write(2) can't
+ * read from the vsyscall page on any kernel version or mode.  The
+ * fact that ptrace() ever worked was a nice courtesy of old kernels,
+ * but the code to support it is fairly gross.
+ */
 static int test_process_vm_readv(void)
 {
 #ifdef __x86_64__

@@ -477,8 +488,12 @@ static int test_process_vm_readv(void)
 	remote.iov_len = 4096;
 	ret = process_vm_readv(getpid(), &local, 1, &remote, 1, 0);
 	if (ret != 4096) {
-		printf("[OK]\tprocess_vm_readv() failed (ret = %d, errno = %d)\n", ret, errno);
-		return 0;
+		/*
+		 * We expect process_vm_readv() to work if and only if the
+		 * vsyscall page is readable.
+		 */
+		printf("[%s]\tprocess_vm_readv() failed (ret = %d, errno = %d)\n", vsyscall_map_r ? "FAIL" : "OK", ret, errno);
+		return vsyscall_map_r ? 1 : 0;
 	}
 
 	if (vsyscall_map_r) {

@@ -488,6 +503,9 @@ static int test_process_vm_readv(void)
 			printf("[FAIL]\tIt worked but returned incorrect data\n");
 			return 1;
 		}
+	} else {
+		printf("[FAIL]\tprocess_rm_readv() succeeded, but it should have failed in this configuration\n");
+		return 1;
 	}
 #endif
 

From: Kim Phillips <[email protected]>

commit e48a73a312ebf19cc3d72aa74985db25c30757c1 upstream.

Event modifiers are not mentioned in the perf record or perf stat manpages. Add them to orient new users more effectively by pointing them to the perf list manpage for details.

---
 tools/perf/Documentation/perf-record.txt |    4 ++++
 tools/perf/Documentation/perf-stat.txt   |    4 ++++
 2 files changed, 8 insertions(+)

@@ -33,6 +33,10 @@ OPTIONS
         - a raw PMU event (eventsel+umask) in the form of rNNN where NNN is a
 	  hexadecimal event descriptor.
 
+        - a symbolic or raw PMU event followed by an optional colon
+	  and a list of event modifiers, e.g., cpu-cycles:p.  See the
+	  linkperf:perf-list[1] man page for details on event modifiers.
+
 	- a symbolically formed PMU event like 'pmu/param1=0x3,param2/' where
 	  'param1', 'param2', etc are defined as formats for the PMU in
 	  /sys/bus/event_source/devices/<pmu>/format/*.

@@ -39,6 +39,10 @@ report::
 	- a raw PMU event (eventsel+umask) in the form of rNNN where NNN is a
 	  hexadecimal event descriptor.
 
+        - a symbolic or raw PMU event followed by an optional colon
+	  and a list of event modifiers, e.g., cpu-cycles:p.  See the
+	  linkperf:perf-list[1] man page for details on event modifiers.
+
 	- a symbolically formed event like 'pmu/param1=0x3,param2/' where
 	  param1 and param2 are defined as formats for the PMU in
 	  /sys/bus/event_source/devices/<pmu>/format/*

From: Peilin Ye <[email protected]>

commit 4ffb879ea648c2b42da4ca992ed3db87e564af69 upstream.

video_put_user() is copying uninitialized stack memory to userspace due to the compiler not initializing holes in the structures declared on the stack. Fix it by initializing ev32 and vb32 using memset().

---
 drivers/media/v4l2-core/v4l2-ioctl.c |   50 ++++++++++++++++++-----------------
 1 file changed, 27 insertions(+), 23 deletions(-)

@@ -3186,14 +3186,16 @@ static int video_put_user(void __user *a
 #ifdef CONFIG_COMPAT_32BIT_TIME
 	case VIDIOC_DQEVENT_TIME32: {
 		struct v4l2_event *ev = parg;
-		struct v4l2_event_time32 ev32 = {
-			.type		= ev->type,
-			.pending	= ev->pending,
-			.sequence	= ev->sequence,
-			.timestamp.tv_sec  = ev->timestamp.tv_sec,
-			.timestamp.tv_nsec = ev->timestamp.tv_nsec,
-			.id		= ev->id,
-		};
+		struct v4l2_event_time32 ev32;
+
+		memset(&ev32, 0, sizeof(ev32));
+
+		ev32.type	= ev->type;
+		ev32.pending	= ev->pending;
+		ev32.sequence	= ev->sequence;
+		ev32.timestamp.tv_sec	= ev->timestamp.tv_sec;
+		ev32.timestamp.tv_nsec	= ev->timestamp.tv_nsec;
+		ev32.id		= ev->id;
 
 		memcpy(&ev32.u, &ev->u, sizeof(ev->u));
 		memcpy(&ev32.reserved, &ev->reserved, sizeof(ev->reserved));

@@ -3207,21 +3209,23 @@ static int video_put_user(void __user *a
 	case VIDIOC_DQBUF_TIME32:
 	case VIDIOC_PREPARE_BUF_TIME32: {
 		struct v4l2_buffer *vb = parg;
-		struct v4l2_buffer_time32 vb32 = {
-			.index		= vb->index,
-			.type		= vb->type,
-			.bytesused	= vb->bytesused,
-			.flags		= vb->flags,
-			.field		= vb->field,
-			.timestamp.tv_sec	= vb->timestamp.tv_sec,
-			.timestamp.tv_usec	= vb->timestamp.tv_usec,
-			.timecode	= vb->timecode,
-			.sequence	= vb->sequence,
-			.memory		= vb->memory,
-			.m.userptr	= vb->m.userptr,
-			.length		= vb->length,
-			.request_fd	= vb->request_fd,
-		};
+		struct v4l2_buffer_time32 vb32;
+
+		memset(&vb32, 0, sizeof(vb32));
+
+		vb32.index	= vb->index;
+		vb32.type	= vb->type;
+		vb32.bytesused	= vb->bytesused;
+		vb32.flags	= vb->flags;
+		vb32.field	= vb->field;
+		vb32.timestamp.tv_sec	= vb->timestamp.tv_sec;
+		vb32.timestamp.tv_usec	= vb->timestamp.tv_usec;
+		vb32.timecode	= vb->timecode;
+		vb32.sequence	= vb->sequence;
+		vb32.memory	= vb->memory;
+		vb32.m.userptr	= vb->m.userptr;
+		vb32.length	= vb->length;
+		vb32.request_fd	= vb->request_fd;
 
 		if (copy_to_user(arg, &vb32, sizeof(vb32)))
 			return -EFAULT;

From: James Morse <[email protected]>

commit e9ee186bb735bfc17fa81dbc9aebf268aee5b41e upstream.

KVM has a one instruction window where it will allow an SError exception to be consumed by the hypervisor without treating it as a hypervisor bug. This is used to consume asynchronous external abort that were caused by the guest.

As we are about to add another location that survives unexpected exceptions, generalise this code to make it behave like the host's extable.

KVM's version has to be mapped to EL2 to be accessible on nVHE systems.

The SError vaxorcism code is a one instruction window, so has two entries in the extable. Because the KVM code is copied for VHE and nVHE, we end up with four entries, half of which correspond with code that isn't mapped.

---
 arch/arm64/include/asm/kvm_asm.h |   15 +++++++++++
 arch/arm64/kernel/vmlinux.lds.S  |    8 ++++++
 arch/arm64/kvm/hyp/entry.S       |   15 ++++++-----
 arch/arm64/kvm/hyp/hyp-entry.S   |   51 +++++++++++++++++++++++----------------
 arch/arm64/kvm/hyp/switch.c      |   31 +++++++++++++++++++++++
 5 files changed, 94 insertions(+), 26 deletions(-)

@@ -146,6 +146,21 @@ extern char __smccc_workaround_1_smc[__S
 	kern_hyp_va	\vcpu
 .endm
 
+/*
+ * KVM extable for unexpected exceptions.
+ * In the same format _asm_extable, but output to a different section so that
+ * it can be mapped to EL2. The KVM version is not sorted. The caller must
+ * ensure:
+ * x18 has the hypervisor value to allow any Shadow-Call-Stack instrumented
+ * code to write to it, and that SPSR_EL2 and ELR_EL2 are restored by the fixup.
+ */
+.macro	_kvm_extable, from, to
+	.pushsection	__kvm_ex_table, "a"
+	.align		3
+	.long		(\from - .), (\to - .)
+	.popsection
+.endm
+
 #endif
 
 #endif /* __ARM_KVM_ASM_H__ */

@@ -21,6 +21,13 @@ ENTRY(_text)
 
 jiffies = jiffies_64;
 
+
+#define HYPERVISOR_EXTABLE					\
+	. = ALIGN(SZ_8);					\
+	__start___kvm_ex_table = .;				\
+	*(__kvm_ex_table)					\
+	__stop___kvm_ex_table = .;
+
 #define HYPERVISOR_TEXT					\
 	/*						\
 	 * Align to 4 KB so that			\

@@ -36,6 +43,7 @@ jiffies = jiffies_64;
 	__hyp_idmap_text_end = .;			\
 	__hyp_text_start = .;				\
 	*(.hyp.text)					\
+	HYPERVISOR_EXTABLE				\
 	__hyp_text_end = .;
 
 #define IDMAP_TEXT					\

@@ -198,20 +198,23 @@ alternative_endif
 	// This is our single instruction exception window. A pending
 	// SError is guaranteed to occur at the earliest when we unmask
 	// it, and at the latest just after the ISB.
-	.global	abort_guest_exit_start
 abort_guest_exit_start:
 
 	isb
 
-	.global	abort_guest_exit_end
 abort_guest_exit_end:
 
 	msr	daifset, #4	// Mask aborts
+	ret
 
-	// If the exception took place, restore the EL1 exception
-	// context so that we can report some information.
-	// Merge the exception code with the SError pending bit.
-	tbz	x0, #ARM_EXIT_WITH_SERROR_BIT, 1f
+	_kvm_extable	abort_guest_exit_start, 9997f
+	_kvm_extable	abort_guest_exit_end, 9997f
+9997:
+	msr	daifset, #4	// Mask aborts
+	mov	x0, #(1 << ARM_EXIT_WITH_SERROR_BIT)
+
+	// restore the EL1 exception context so that we can report some
+	// information. Merge the exception code with the SError pending bit.
 	msr	elr_el2, x2
 	msr	esr_el2, x3
 	msr	spsr_el2, x4

@@ -14,6 +14,7 @@
 
 #include <asm/barrier.h>
 #include <asm/cpufeature.h>
+#include <asm/extable.h>
 #include <asm/kprobes.h>
 #include <asm/kvm_asm.h>
 #include <asm/kvm_emulate.h>

@@ -24,6 +25,9 @@
 #include <asm/processor.h>
 #include <asm/thread_info.h>
 
+extern struct exception_table_entry __start___kvm_ex_table;
+extern struct exception_table_entry __stop___kvm_ex_table;
+
 /* Check whether the FP regs were dirtied while in the host-side run loop: */
 static bool __hyp_text update_fp_enabled(struct kvm_vcpu *vcpu)
 {

@@ -934,3 +938,30 @@ void __hyp_text __noreturn hyp_panic(str
 
 	unreachable();
 }
+
+asmlinkage void __hyp_text kvm_unexpected_el2_exception(void)
+{
+	unsigned long addr, fixup;
+	struct kvm_cpu_context *host_ctxt;
+	struct exception_table_entry *entry, *end;
+	unsigned long elr_el2 = read_sysreg(elr_el2);
+
+	entry = hyp_symbol_addr(__start___kvm_ex_table);
+	end = hyp_symbol_addr(__stop___kvm_ex_table);
+	host_ctxt = &__hyp_this_cpu_ptr(kvm_host_data)->host_ctxt;
+
+	while (entry < end) {
+		addr = (unsigned long)&entry->insn + entry->insn;
+		fixup = (unsigned long)&entry->fixup + entry->fixup;
+
+		if (addr != elr_el2) {
+			entry++;
+			continue;
+		}
+
+		write_sysreg(fixup, elr_el2);
+		return;
+	}
+
+	hyp_panic(host_ctxt);
+}

From: James Morse <[email protected]>

commit 88a84ccccb3966bcc3f309cdb76092a9892c0260 upstream.

KVM doesn't expect any synchronous exceptions when executing, any such exception leads to a panic(). AT instructions access the guest page tables, and can cause a synchronous external abort to be taken.

The arm-arm is unclear on what should happen if the guest has configured the hardware update of the access-flag, and a memory type in TCR_EL1 that does not support atomic operations. B2.2.6 "Possible implementation restrictions on using atomic instructions" from DDI0487F.a lists synchronous external abort as a possible behaviour of atomic instructions that target memory that isn't writeback cacheable, but the page table walker may behave differently.

Make KVM robust to synchronous exceptions caused by AT instructions. Add a get_user() style helper for AT instructions that returns -EFAULT if an exception was generated.

While KVM's version of the exception table mixes synchronous and asynchronous exceptions, only one of these can occur at each location.

Re-enter the guest when the AT instructions take an exception on the assumption the guest will take the same exception. This isn't guaranteed to make forward progress, as the AT instructions may always walk the page tables, but guest execution may use the translation cached in the TLB.

This isn't a problem, as since commit 5dcd0fdbb492 ("KVM: arm64: Defer guest entry when an asynchronous exception is pending"), KVM will return to the host to process IRQs allowing the rest of the system to keep running.

---
 arch/arm64/include/asm/kvm_asm.h |   28 ++++++++++++++++++++++++++++
 arch/arm64/kvm/hyp/hyp-entry.S   |   14 ++++++++++----
 arch/arm64/kvm/hyp/switch.c      |    8 ++++----
 3 files changed, 42 insertions(+), 8 deletions(-)

@@ -121,6 +121,34 @@ extern char __smccc_workaround_1_smc[__S
 		*__hyp_this_cpu_ptr(sym);				\
 	 })
 
+#define __KVM_EXTABLE(from, to)						\
+	"	.pushsection	__kvm_ex_table, \"a\"\n"		\
+	"	.align		3\n"					\
+	"	.long		(" #from " - .), (" #to " - .)\n"	\
+	"	.popsection\n"
+
+
+#define __kvm_at(at_op, addr)						\
+( { 									\
+	int __kvm_at_err = 0;						\
+	u64 spsr, elr;							\
+	asm volatile(							\
+	"	mrs	%1, spsr_el2\n"					\
+	"	mrs	%2, elr_el2\n"					\
+	"1:	at	"at_op", %3\n"					\
+	"	isb\n"							\
+	"	b	9f\n"						\
+	"2:	msr	spsr_el2, %1\n"					\
+	"	msr	elr_el2, %2\n"					\
+	"	mov	%w0, %4\n"					\
+	"9:\n"								\
+	__KVM_EXTABLE(1b, 2b)						\
+	: "+r" (__kvm_at_err), "=&r" (spsr), "=&r" (elr)		\
+	: "r" (addr), "i" (-EFAULT));					\
+	__kvm_at_err;							\
+} )
+
+
 #else /* __ASSEMBLY__ */
 
 .macro hyp_adr_this_cpu reg, sym, tmp

@@ -166,13 +166,19 @@ el1_error:
 	b	__guest_exit
 
 el2_sync:
-	/* Check for illegal exception return, otherwise panic */
+	/* Check for illegal exception return */
 	mrs	x0, spsr_el2
+	tbnz	x0, #20, 1f
 
-	/* if this was something else, then panic! */
-	tst	x0, #PSR_IL_BIT
-	b.eq	__hyp_panic
+	save_caller_saved_regs_vect
+	stp     x29, x30, [sp, #-16]!
+	bl	kvm_unexpected_el2_exception
+	ldp     x29, x30, [sp], #16
+	restore_caller_saved_regs_vect
 
+	eret
+
+1:
 	/* Let's attempt a recovery from the illegal exception return */
 	get_vcpu_ptr	x1, x0
 	mov	x0, #ARM_EXCEPTION_IL

@@ -303,10 +303,10 @@ static bool __hyp_text __translate_far_t
 	 * saved the guest context yet, and we may return early...
 	 */
 	par = read_sysreg(par_el1);
-	asm volatile("at s1e1r, %0" : : "r" (far));
-	isb();
-
-	tmp = read_sysreg(par_el1);
+	if (!__kvm_at("s1e1r", far))
+		tmp = read_sysreg(par_el1);
+	else
+		tmp = SYS_PAR_EL1_F; /* back to the guest */
 	write_sysreg(par, par_el1);
 
 	if (unlikely(tmp & SYS_PAR_EL1_F))

From: Sowjanya Komatineni <[email protected]>

commit f7f86e8ac0ad7cd6792a80137f5a550924966916 upstream.

commit b5a84ecf025a ("mmc: tegra: Add Tegra210 support")

Tegra210 and later uses separate SDMMC_LEGACY_TM clock for data timeout.

So, this patch adds "tmclk" to Tegra sdhci clock property in the device tree binding.

---
 Documentation/devicetree/bindings/mmc/nvidia,tegra20-sdhci.txt |   32 +++++++++-
 1 file changed, 29 insertions(+), 3 deletions(-)

@@ -15,8 +15,15 @@ Required properties:
   - "nvidia,tegra210-sdhci": for Tegra210
   - "nvidia,tegra186-sdhci": for Tegra186
   - "nvidia,tegra194-sdhci": for Tegra194
-- clocks : Must contain one entry, for the module clock.
-  See ../clocks/clock-bindings.txt for details.
+- clocks: For Tegra210, Tegra186 and Tegra194 must contain two entries.
+	  One for the module clock and one for the timeout clock.
+	  For all other Tegra devices, must contain a single entry for
+	  the module clock. See ../clocks/clock-bindings.txt for details.
+- clock-names: For Tegra210, Tegra186 and Tegra194 must contain the
+	       strings 'sdhci' and 'tmclk' to represent the module and
+	       the timeout clocks, respectively.
+	       For all other Tegra devices must contain the string 'sdhci'
+	       to represent the module clock.
 - resets : Must contain an entry for each entry in reset-names.
   See ../reset/reset.txt for details.
 - reset-names : Must include the following entries:

@@ -99,7 +106,7 @@ Optional properties for Tegra210, Tegra1
 
 Example:
 sdhci@700b0000 {
-	compatible = "nvidia,tegra210-sdhci", "nvidia,tegra124-sdhci";
+	compatible = "nvidia,tegra124-sdhci";
 	reg = <0x0 0x700b0000 0x0 0x200>;
 	interrupts = <GIC_SPI 14 IRQ_TYPE_LEVEL_HIGH>;
 	clocks = <&tegra_car TEGRA210_CLK_SDMMC1>;

@@ -107,6 +114,25 @@ sdhci@700b0000 {
 	resets = <&tegra_car 14>;
 	reset-names = "sdhci";
 	pinctrl-names = "sdmmc-3v3", "sdmmc-1v8";
+	pinctrl-0 = <&sdmmc1_3v3>;
+	pinctrl-1 = <&sdmmc1_1v8>;
+	nvidia,pad-autocal-pull-up-offset-3v3 = <0x00>;
+	nvidia,pad-autocal-pull-down-offset-3v3 = <0x7d>;
+	nvidia,pad-autocal-pull-up-offset-1v8 = <0x7b>;
+	nvidia,pad-autocal-pull-down-offset-1v8 = <0x7b>;
+	status = "disabled";
+};
+
+sdhci@700b0000 {
+	compatible = "nvidia,tegra210-sdhci";
+	reg = <0x0 0x700b0000 0x0 0x200>;
+	interrupts = <GIC_SPI 14 IRQ_TYPE_LEVEL_HIGH>;
+	clocks = <&tegra_car TEGRA210_CLK_SDMMC1>,
+		 <&tegra_car TEGRA210_CLK_SDMMC_LEGACY>;
+	clock-names = "sdhci", "tmclk";
+	resets = <&tegra_car 14>;
+	reset-names = "sdhci";
+	pinctrl-names = "sdmmc-3v3", "sdmmc-1v8";
 	pinctrl-0 = <&sdmmc1_3v3>;
 	pinctrl-1 = <&sdmmc1_1v8>;
 	nvidia,pad-autocal-pull-up-offset-3v3 = <0x00>;

From: Sowjanya Komatineni <[email protected]>

commit c956c0cd4f6f4aac4f095621b1c4e1c5ee1df877 upstream.

commit 5425fb15d8ee ("arm64: tegra: Add Tegra194 chip device tree")

Tegra194 uses separate SDMMC_LEGACY_TM clock for data timeout and this clock is not enabled currently which is not recommended.

Tegra194 SDMMC advertises 12Mhz as timeout clock frequency in host capability register.

So, this clock should be kept enabled by SDMMC driver.

---
 arch/arm64/boot/dts/nvidia/tegra194.dtsi |   15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

@@ -453,8 +453,9 @@
 			compatible = "nvidia,tegra194-sdhci", "nvidia,tegra186-sdhci";
 			reg = <0x03400000 0x10000>;
 			interrupts = <GIC_SPI 62 IRQ_TYPE_LEVEL_HIGH>;
-			clocks = <&bpmp TEGRA194_CLK_SDMMC1>;
-			clock-names = "sdhci";
+			clocks = <&bpmp TEGRA194_CLK_SDMMC1>,
+				 <&bpmp TEGRA194_CLK_SDMMC_LEGACY_TM>;
+			clock-names = "sdhci", "tmclk";
 			resets = <&bpmp TEGRA194_RESET_SDMMC1>;
 			reset-names = "sdhci";
 			nvidia,pad-autocal-pull-up-offset-3v3-timeout =

@@ -475,8 +476,9 @@
 			compatible = "nvidia,tegra194-sdhci", "nvidia,tegra186-sdhci";
 			reg = <0x03440000 0x10000>;
 			interrupts = <GIC_SPI 64 IRQ_TYPE_LEVEL_HIGH>;
-			clocks = <&bpmp TEGRA194_CLK_SDMMC3>;
-			clock-names = "sdhci";
+			clocks = <&bpmp TEGRA194_CLK_SDMMC3>,
+				 <&bpmp TEGRA194_CLK_SDMMC_LEGACY_TM>;
+			clock-names = "sdhci", "tmclk";
 			resets = <&bpmp TEGRA194_RESET_SDMMC3>;
 			reset-names = "sdhci";
 			nvidia,pad-autocal-pull-up-offset-1v8 = <0x00>;

@@ -498,8 +500,9 @@
 			compatible = "nvidia,tegra194-sdhci", "nvidia,tegra186-sdhci";
 			reg = <0x03460000 0x10000>;
 			interrupts = <GIC_SPI 65 IRQ_TYPE_LEVEL_HIGH>;
-			clocks = <&bpmp TEGRA194_CLK_SDMMC4>;
-			clock-names = "sdhci";
+			clocks = <&bpmp TEGRA194_CLK_SDMMC4>,
+				 <&bpmp TEGRA194_CLK_SDMMC_LEGACY_TM>;
+			clock-names = "sdhci", "tmclk";
 			assigned-clocks = <&bpmp TEGRA194_CLK_SDMMC4>,
 					  <&bpmp TEGRA194_CLK_PLLC4>;
 			assigned-clock-parents =

From: Sowjanya Komatineni <[email protected]>

commit baba217d2c4446b6eef309d81d8776cb5c68cb55 upstream.

commit 39cb62cb8973 ("arm64: tegra: Add Tegra186 support")

Tegra186 uses separate SDMMC_LEGACY_TM clock for data timeout and this clock is not enabled currently which is not recommended.

Tegra186 SDMMC advertises 12Mhz as timeout clock frequency in host capability register and uses it by default.

So, this clock should be kept enabled by the SDMMC driver.

---
 arch/arm64/boot/dts/nvidia/tegra186.dtsi |   20 ++++++++++++--------
 1 file changed, 12 insertions(+), 8 deletions(-)

@@ -331,8 +331,9 @@
 		compatible = "nvidia,tegra186-sdhci";
 		reg = <0x0 0x03400000 0x0 0x10000>;
 		interrupts = <GIC_SPI 62 IRQ_TYPE_LEVEL_HIGH>;
-		clocks = <&bpmp TEGRA186_CLK_SDMMC1>;
-		clock-names = "sdhci";
+		clocks = <&bpmp TEGRA186_CLK_SDMMC1>,
+			 <&bpmp TEGRA186_CLK_SDMMC_LEGACY_TM>;
+		clock-names = "sdhci", "tmclk";
 		resets = <&bpmp TEGRA186_RESET_SDMMC1>;
 		reset-names = "sdhci";
 		iommus = <&smmu TEGRA186_SID_SDMMC1>;

@@ -357,8 +358,9 @@
 		compatible = "nvidia,tegra186-sdhci";
 		reg = <0x0 0x03420000 0x0 0x10000>;
 		interrupts = <GIC_SPI 63 IRQ_TYPE_LEVEL_HIGH>;
-		clocks = <&bpmp TEGRA186_CLK_SDMMC2>;
-		clock-names = "sdhci";
+		clocks = <&bpmp TEGRA186_CLK_SDMMC2>,
+			 <&bpmp TEGRA186_CLK_SDMMC_LEGACY_TM>;
+		clock-names = "sdhci", "tmclk";
 		resets = <&bpmp TEGRA186_RESET_SDMMC2>;
 		reset-names = "sdhci";
 		iommus = <&smmu TEGRA186_SID_SDMMC2>;

@@ -378,8 +380,9 @@
 		compatible = "nvidia,tegra186-sdhci";
 		reg = <0x0 0x03440000 0x0 0x10000>;
 		interrupts = <GIC_SPI 64 IRQ_TYPE_LEVEL_HIGH>;
-		clocks = <&bpmp TEGRA186_CLK_SDMMC3>;
-		clock-names = "sdhci";
+		clocks = <&bpmp TEGRA186_CLK_SDMMC3>,
+			 <&bpmp TEGRA186_CLK_SDMMC_LEGACY_TM>;
+		clock-names = "sdhci", "tmclk";
 		resets = <&bpmp TEGRA186_RESET_SDMMC3>;
 		reset-names = "sdhci";
 		iommus = <&smmu TEGRA186_SID_SDMMC3>;

@@ -401,8 +404,9 @@
 		compatible = "nvidia,tegra186-sdhci";
 		reg = <0x0 0x03460000 0x0 0x10000>;
 		interrupts = <GIC_SPI 65 IRQ_TYPE_LEVEL_HIGH>;
-		clocks = <&bpmp TEGRA186_CLK_SDMMC4>;
-		clock-names = "sdhci";
+		clocks = <&bpmp TEGRA186_CLK_SDMMC4>,
+			 <&bpmp TEGRA186_CLK_SDMMC_LEGACY_TM>;
+		clock-names = "sdhci", "tmclk";
 		assigned-clocks = <&bpmp TEGRA186_CLK_SDMMC4>,
 				  <&bpmp TEGRA186_CLK_PLLC4_VCO>;
 		assigned-clock-parents = <&bpmp TEGRA186_CLK_PLLC4_VCO>;

From: Sowjanya Komatineni <[email protected]>

commit 679f71fa0db2d777f39c7a5af7f7c0689fc713fa upstream.

commit 742af7e7a0a1 ("arm64: tegra: Add Tegra210 support")

Tegra210 uses separate SDMMC_LEGACY_TM clock for data timeout and this clock is not enabled currently which is not recommended.

Tegra SDMMC advertises 12Mhz as timeout clock frequency in host capability register.

So, this clock should be kept enabled by SDMMC driver.

---
 arch/arm64/boot/dts/nvidia/tegra210.dtsi |   20 ++++++++++++--------
 1 file changed, 12 insertions(+), 8 deletions(-)

@@ -1180,8 +1180,9 @@
 		compatible = "nvidia,tegra210-sdhci", "nvidia,tegra124-sdhci";
 		reg = <0x0 0x700b0000 0x0 0x200>;
 		interrupts = <GIC_SPI 14 IRQ_TYPE_LEVEL_HIGH>;
-		clocks = <&tegra_car TEGRA210_CLK_SDMMC1>;
-		clock-names = "sdhci";
+		clocks = <&tegra_car TEGRA210_CLK_SDMMC1>,
+			 <&tegra_car TEGRA210_CLK_SDMMC_LEGACY>;
+		clock-names = "sdhci", "tmclk";
 		resets = <&tegra_car 14>;
 		reset-names = "sdhci";
 		pinctrl-names = "sdmmc-3v3", "sdmmc-1v8",

@@ -1208,8 +1209,9 @@
 		compatible = "nvidia,tegra210-sdhci", "nvidia,tegra124-sdhci";
 		reg = <0x0 0x700b0200 0x0 0x200>;
 		interrupts = <GIC_SPI 15 IRQ_TYPE_LEVEL_HIGH>;
-		clocks = <&tegra_car TEGRA210_CLK_SDMMC2>;
-		clock-names = "sdhci";
+		clocks = <&tegra_car TEGRA210_CLK_SDMMC2>,
+			 <&tegra_car TEGRA210_CLK_SDMMC_LEGACY>;
+		clock-names = "sdhci", "tmclk";
 		resets = <&tegra_car 9>;
 		reset-names = "sdhci";
 		pinctrl-names = "sdmmc-1v8-drv";

@@ -1225,8 +1227,9 @@
 		compatible = "nvidia,tegra210-sdhci", "nvidia,tegra124-sdhci";
 		reg = <0x0 0x700b0400 0x0 0x200>;
 		interrupts = <GIC_SPI 19 IRQ_TYPE_LEVEL_HIGH>;
-		clocks = <&tegra_car TEGRA210_CLK_SDMMC3>;
-		clock-names = "sdhci";
+		clocks = <&tegra_car TEGRA210_CLK_SDMMC3>,
+			 <&tegra_car TEGRA210_CLK_SDMMC_LEGACY>;
+		clock-names = "sdhci", "tmclk";
 		resets = <&tegra_car 69>;
 		reset-names = "sdhci";
 		pinctrl-names = "sdmmc-3v3", "sdmmc-1v8",

@@ -1248,8 +1251,9 @@
 		compatible = "nvidia,tegra210-sdhci", "nvidia,tegra124-sdhci";
 		reg = <0x0 0x700b0600 0x0 0x200>;
 		interrupts = <GIC_SPI 31 IRQ_TYPE_LEVEL_HIGH>;
-		clocks = <&tegra_car TEGRA210_CLK_SDMMC4>;
-		clock-names = "sdhci";
+		clocks = <&tegra_car TEGRA210_CLK_SDMMC4>,
+			 <&tegra_car TEGRA210_CLK_SDMMC_LEGACY>;
+		clock-names = "sdhci", "tmclk";
 		resets = <&tegra_car 15>;
 		reset-names = "sdhci";
 		pinctrl-names = "sdmmc-3v3-drv", "sdmmc-1v8-drv";

From: Sowjanya Komatineni <[email protected]>

commit e33588adcaa925c18ee2ea253161fb0317fa2329 upstream.

commit b5a84ecf025a ("mmc: tegra: Add Tegra210 support")

SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK is set for Tegra210 from the beginning of Tegra210 support in the driver.

Tegra210 SDMMC hardware by default uses timeout clock (TMCLK) instead of SDCLK and this quirk should not be set.

So, this patch remove this quirk for Tegra210.

---
 drivers/mmc/host/sdhci-tegra.c |    1 -
 1 file changed, 1 deletion(-)

@@ -1409,7 +1409,6 @@ static const struct sdhci_ops tegra210_s
 
 static const struct sdhci_pltfm_data sdhci_tegra210_pdata = {
 	.quirks = SDHCI_QUIRK_BROKEN_TIMEOUT_VAL |
-		  SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK |
 		  SDHCI_QUIRK_SINGLE_POWER_WRITE |
 		  SDHCI_QUIRK_NO_HISPD_BIT |
 		  SDHCI_QUIRK_BROKEN_ADMA_ZEROLEN_DESC |

From: Sowjanya Komatineni <[email protected]>

commit 391d89dba8c290859a3e29430d0b9e32c358bb0d upstream.

commit 4346b7c7941d ("mmc: tegra: Add Tegra186 support")

SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK is set for Tegra186 from the beginning of its support in driver.

Tegra186 SDMMC hardware by default uses timeout clock (TMCLK) instead of SDCLK and this quirk should not be set.

So, this patch remove this quirk for Tegra186.

---
 drivers/mmc/host/sdhci-tegra.c |    1 -
 1 file changed, 1 deletion(-)

@@ -1446,7 +1446,6 @@ static const struct sdhci_ops tegra186_s
 
 static const struct sdhci_pltfm_data sdhci_tegra186_pdata = {
 	.quirks = SDHCI_QUIRK_BROKEN_TIMEOUT_VAL |
-		  SDHCI_QUIRK_DATA_TIMEOUT_USES_SDCLK |
 		  SDHCI_QUIRK_SINGLE_POWER_WRITE |
 		  SDHCI_QUIRK_NO_HISPD_BIT |
 		  SDHCI_QUIRK_BROKEN_ADMA_ZEROLEN_DESC |

From: Johannes Berg <[email protected]>

commit fce2ff728f95b8894db14f51c9274dc56c37616f upstream.

In nl80211_set_station(), we check NL80211_ATTR_HE_6GHZ_CAPABILITY and then use NL80211_ATTR_HE_CAPABILITY, which is clearly wrong. Fix this to use NL80211_ATTR_HE_6GHZ_CAPABILITY as well.

---
 net/wireless/nl80211.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

@@ -6010,7 +6010,7 @@ static int nl80211_set_station(struct sk
 
 	if (info->attrs[NL80211_ATTR_HE_6GHZ_CAPABILITY])
 		params.he_6ghz_capa =
-			nla_data(info->attrs[NL80211_ATTR_HE_CAPABILITY]);
+			nla_data(info->attrs[NL80211_ATTR_HE_6GHZ_CAPABILITY]);
 
 	if (info->attrs[NL80211_ATTR_AIRTIME_WEIGHT])
 		params.airtime_weight =

From: Bodo Stroesser <[email protected]>

commit 3c58f737231e2c8cbf543a09d84d8c8e80e05e43 upstream.

(scatter|gather)_data_area() need to flush dcache after writing data to or before reading data from a page in uio data area. The two routines are able to handle data transfer to/from such a page in fragments and flush the cache after each fragment was copied by calling the wrapper tcmu_flush_dcache_range().

That means:

1) flush_dcache_page() can be called multiple times for the same page.

2) Calling flush_dcache_page() indirectly using the wrapper does not make sense, because each call of the wrapper is for one single page only and the calling routine already has the correct page pointer.

Change (scatter|gather)_data_area() such that, instead of calling tcmu_flush_dcache_range() before/after each memcpy, it now calls flush_dcache_page() before unmapping a page (when writing is complete for that page) or after mapping a page (when starting to read the page).

After this change only calls to tcmu_flush_dcache_range() for addresses in vmalloc'ed command ring are left over.

The patch was tested on ARM with kernel 4.19.118 and 5.7.2

---
 drivers/target/target_core_user.c |   11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

@@ -676,8 +676,10 @@ static void scatter_data_area(struct tcm
 		from = kmap_atomic(sg_page(sg)) + sg->offset;
 		while (sg_remaining > 0) {
 			if (block_remaining == 0) {
-				if (to)
+				if (to) {
+					flush_dcache_page(page);
 					kunmap_atomic(to);
+				}
 
 				block_remaining = DATA_BLOCK_SIZE;
 				dbi = tcmu_cmd_get_dbi(tcmu_cmd);

@@ -722,7 +724,6 @@ static void scatter_data_area(struct tcm
 				memcpy(to + offset,
 				       from + sg->length - sg_remaining,
 				       copy_bytes);
-				tcmu_flush_dcache_range(to, copy_bytes);
 			}
 
 			sg_remaining -= copy_bytes;

@@ -731,8 +732,10 @@ static void scatter_data_area(struct tcm
 		kunmap_atomic(from - sg->offset);
 	}
 
-	if (to)
+	if (to) {
+		flush_dcache_page(page);
 		kunmap_atomic(to);
+	}
 }
 
 static void gather_data_area(struct tcmu_dev *udev, struct tcmu_cmd *cmd,

@@ -778,13 +781,13 @@ static void gather_data_area(struct tcmu
 				dbi = tcmu_cmd_get_dbi(cmd);
 				page = tcmu_get_block_page(udev, dbi);
 				from = kmap_atomic(page);
+				flush_dcache_page(page);
 			}
 			copy_bytes = min_t(size_t, sg_remaining,
 					block_remaining);
 			if (read_len < copy_bytes)
 				copy_bytes = read_len;
 			offset = DATA_BLOCK_SIZE - block_remaining;
-			tcmu_flush_dcache_range(from, copy_bytes);
 			memcpy(to + sg->length - sg_remaining, from + offset,
 					copy_bytes);
 

Build results:

	total: 154 pass: 154 fail: 0
Qemu test results:
	total: 430 pass: 430 fail: 0

Guenter

Thanks for testing both of these and letting me know.

greg k-h

Compiled and booted on my test system. No dmesg regressions.

thanks,
-- Shuah

Thanks for testing both of these and letting me know.

greg k-h

Results from Linaro’s test farm.
No regressions on arm64, arm, x86_64, and i386.

Summary
------------------------------------------------------------------------

kernel: 5.8.7-rc1
git repo: https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable-rc.git
git branch: linux-5.8.y
git commit: 072d7559faf2a3b9f3c93e5cc96b276c6f02adec
git describe: v5.8.5-272-g072d7559faf2
Test details: https://qa-reports.linaro.org/lkft/linux-stable-rc-5.8-oe/build/v5.8.5-272-g072d7559faf2

No regressions (compared to build v5.8.5)

No fixes (compared to build v5.8.5)

Ran 35624 total tests in the following environments and test suites.

Environments
--------------
- dragonboard-410c
- hi6220-hikey
- i386
- juno-r2
- juno-r2-compat
- juno-r2-kasan
- nxp-ls2088
- qemu_arm
- qemu_arm64
- qemu_i386
- qemu_x86_64
- x15
- x86
- x86-kasan

Test Suites
-----------
* igt-gpu-tools
* libhugetlbfs
* linux-log-parser
* ltp-cap_bounds-tests
* ltp-commands-tests
* ltp-containers-tests
* ltp-controllers-tests
* ltp-cpuhotplug-tests
* ltp-crypto-tests
* ltp-dio-tests
* ltp-fcntl-locktests-tests
* ltp-filecaps-tests
* ltp-fs-tests
* ltp-fs_bind-tests
* ltp-fs_perms_simple-tests
* ltp-fsx-tests
* ltp-hugetlb-tests
* ltp-io-tests
* ltp-ipc-tests
* ltp-math-tests
* ltp-mm-tests
* ltp-nptl-tests
* ltp-pty-tests
* ltp-securebits-tests
* ltp-syscalls-tests
* ltp-tracing-tests
* ltp-cve-tests
* ltp-sched-tests
* v4l2-compliance
* ltp-open-posix-tests
* network-basic-tests

--
Linaro LKFT
https://lkft.linaro.org