May 2025 - Linux-stable-mirror

[PATCH] LoongArch: Save and restore CSR.CNTC for hibernation

by Huacai Chen

Save and restore CSR.CNTC for hibernation which is similar to suspend. For host this is unnecessary because sched clock is ensured continuous, but for kvm guest sched clock isn't enough because rdtime.d should also be continuous. Host::rdtime.d = Host::CSR.CNTC + counter Guest::rdtime.d = Host::CSR.CNTC + Host::CSR.GCNTC + Guest::CSR.CNTC + counter so, Guest::rdtime.d = Host::rdtime.d + Host::CSR.GCNTC + Guest::CSR.CNTC To ensure Guest::rdtime.d continuous, Host::rdtime.d should be at first continuous, while Host::CSR.GCNTC / Guest::CSR.CNTC is maintained by KVM. Cc: stable(a)vger.kernel.org Signed-off-by: Xianglai Li <lixianglai(a)loongson.cn> Signed-off-by: Huacai Chen <chenhuacai(a)loongson.cn> --- arch/loongarch/kernel/time.c | 2 +- arch/loongarch/power/hibernate.c | 3 +++ 2 files changed, 4 insertions(+), 1 deletion(-) diff --git a/arch/loongarch/kernel/time.c b/arch/loongarch/kernel/time.c index e2d3bfeb6366..bc75a3a69fc8 100644 --- a/arch/loongarch/kernel/time.c +++ b/arch/loongarch/kernel/time.c @@ -111,7 +111,7 @@ static unsigned long __init get_loops_per_jiffy(void) return lpj; } -static long init_offset __nosavedata; +static long init_offset; void save_counter(void) { diff --git a/arch/loongarch/power/hibernate.c b/arch/loongarch/power/hibernate.c index 1e0590542f98..e7b7346592cb 100644 --- a/arch/loongarch/power/hibernate.c +++ b/arch/loongarch/power/hibernate.c @@ -2,6 +2,7 @@ #include <asm/fpu.h> #include <asm/loongson.h> #include <asm/sections.h> +#include <asm/time.h> #include <asm/tlbflush.h> #include <linux/suspend.h> @@ -14,6 +15,7 @@ struct pt_regs saved_regs; void save_processor_state(void) { + save_counter(); saved_crmd = csr_read32(LOONGARCH_CSR_CRMD); saved_prmd = csr_read32(LOONGARCH_CSR_PRMD); saved_euen = csr_read32(LOONGARCH_CSR_EUEN); @@ -26,6 +28,7 @@ void save_processor_state(void) void restore_processor_state(void) { + sync_counter(); csr_write32(saved_crmd, LOONGARCH_CSR_CRMD); csr_write32(saved_prmd, LOONGARCH_CSR_PRMD); csr_write32(saved_euen, LOONGARCH_CSR_EUEN); -- 2.47.1

3 months, 4 weeks

2
1
0 0

[PATCH v2] usb: cdnsp: Fix issue with detecting command completion event

by Pawel Laszczak

In some cases, there is a small-time gap in which CMD_RING_BUSY can be cleared by controller but adding command completion event to event ring will be delayed. As the result driver will return error code. This behavior has been detected on usbtest driver (test 9) with configuration including ep1in/ep1out bulk and ep2in/ep2out isoc endpoint. Probably this gap occurred because controller was busy with adding some other events to event ring. The CMD_RING_BUSY is cleared to '0' when the Command Descriptor has been executed and not when command completion event has been added to event ring. To fix this issue for this test the small delay is sufficient less than 10us) but to make sure the problem doesn't happen again in the future the patch introduces 10 retries to check with delay about 20us before returning error code. Fixes: 3d82904559f4 ("usb: cdnsp: cdns3 Add main part of Cadence USBSSP DRD Driver") cc: stable(a)vger.kernel.org Signed-off-by: Pawel Laszczak <pawell(a)cadence.com> --- Changelog: v2: - replaced usleep_range with udelay - increased retry counter and decreased the udelay value drivers/usb/cdns3/cdnsp-gadget.c | 18 +++++++++++++++++- 1 file changed, 17 insertions(+), 1 deletion(-) diff --git a/drivers/usb/cdns3/cdnsp-gadget.c b/drivers/usb/cdns3/cdnsp-gadget.c index 4824a10df07e..58650b7f4173 100644 --- a/drivers/usb/cdns3/cdnsp-gadget.c +++ b/drivers/usb/cdns3/cdnsp-gadget.c @@ -547,6 +547,7 @@ int cdnsp_wait_for_cmd_compl(struct cdnsp_device *pdev) dma_addr_t cmd_deq_dma; union cdnsp_trb *event; u32 cycle_state; + u32 retry = 10; int ret, val; u64 cmd_dma; u32 flags; @@ -578,8 +579,23 @@ int cdnsp_wait_for_cmd_compl(struct cdnsp_device *pdev) flags = le32_to_cpu(event->event_cmd.flags); /* Check the owner of the TRB. */ - if ((flags & TRB_CYCLE) != cycle_state) + if ((flags & TRB_CYCLE) != cycle_state) { + /* + * Give some extra time to get chance controller + * to finish command before returning error code. + * Checking CMD_RING_BUSY is not sufficient because + * this bit is cleared to '0' when the Command + * Descriptor has been executed by controller + * and not when command completion event has + * be added to event ring. + */ + if (retry--) { + udelay(20); + continue; + } + return -EINVAL; + } cmd_dma = le64_to_cpu(event->event_cmd.cmd_trb); -- 2.43.0

3 months, 4 weeks

2
3
0 0

[PATCH 1/2] drm/xe/vm: move rebind_work init earlier

by Matthew Auld

In xe_vm_close_and_put() we need to be able to call flush_work(rebind_work), however during vm creation we can call this on the error path, before having actually set up the worker, leading to a splat from flush_work(). It looks like we can simply move the worker init step earlier to fix this. Fixes: dd08ebf6c352 ("drm/xe: Introduce a new DRM driver for Intel GPUs") Signed-off-by: Matthew Auld <matthew.auld(a)intel.com> Cc: Matthew Brost <matthew.brost(a)intel.com> Cc: <stable(a)vger.kernel.org> # v6.8+ --- drivers/gpu/drm/xe/xe_vm.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c index 5a978da411b0..168756fb140b 100644 --- a/drivers/gpu/drm/xe/xe_vm.c +++ b/drivers/gpu/drm/xe/xe_vm.c @@ -1704,8 +1704,10 @@ struct xe_vm *xe_vm_create(struct xe_device *xe, u32 flags) * scheduler drops all the references of it, hence protecting the VM * for this case is necessary. */ - if (flags & XE_VM_FLAG_LR_MODE) + if (flags & XE_VM_FLAG_LR_MODE) { + INIT_WORK(&vm->preempt.rebind_work, preempt_rebind_work_func); xe_pm_runtime_get_noresume(xe); + } vm_resv_obj = drm_gpuvm_resv_object_alloc(&xe->drm); if (!vm_resv_obj) { @@ -1750,10 +1752,8 @@ struct xe_vm *xe_vm_create(struct xe_device *xe, u32 flags) vm->batch_invalidate_tlb = true; } - if (vm->flags & XE_VM_FLAG_LR_MODE) { - INIT_WORK(&vm->preempt.rebind_work, preempt_rebind_work_func); + if (vm->flags & XE_VM_FLAG_LR_MODE) vm->batch_invalidate_tlb = false; - } /* Fill pt_root after allocating scratch tables */ for_each_tile(tile, xe, id) { -- 2.49.0

3 months, 4 weeks

2
1
0 0

[PATCH 0/2] PCI: of: Fix OF device node refcount leakages

by Zijun Hu

This patch series is to fix OF device node refcount leakage for - of_irq_parse_and_map_pci() - of_pci_prop_intr_map() Signed-off-by: Zijun Hu <quic_zijuhu(a)quicinc.com> --- Zijun Hu (2): PCI: of: Fix OF device node refcount leakage in API of_irq_parse_and_map_pci() PCI: of: Fix OF device node refcount leakages in of_pci_prop_intr_map() drivers/pci/of.c | 2 ++ drivers/pci/of_property.c | 20 +++++++++++--------- 2 files changed, 13 insertions(+), 9 deletions(-) --- base-commit: 7d06015d936c861160803e020f68f413b5c3cd9d change-id: 20250407-fix_of_pci-20b45dcc26b5 Best regards, -- Zijun Hu <quic_zijuhu(a)quicinc.com>

3 months, 4 weeks

1
2
0 0

+ highmem-add-folio_test_partial_kmap.patch added to mm-hotfixes-unstable branch

by Andrew Morton

The patch titled Subject: highmem: add folio_test_partial_kmap() has been added to the -mm mm-hotfixes-unstable branch. Its filename is highmem-add-folio_test_partial_kmap.patch This patch will shortly appear at https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patche… This patch will later appear in the mm-hotfixes-unstable branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm Before you just go and hit "reply", please: a) Consider who else should be cc'ed b) Prefer to cc a suitable mailing list as well c) Ideally: find the original patch on the mailing list and do a reply-to-all to that, adding suitable additional cc's *** Remember to use Documentation/process/submit-checklist.rst when testing your code *** The -mm tree is included into linux-next via the mm-everything branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm and is updated there every 2-3 working days ------------------------------------------------------ From: "Matthew Wilcox (Oracle)" <willy(a)infradead.org> Subject: highmem: add folio_test_partial_kmap() Date: Wed, 14 May 2025 18:06:02 +0100 In commit c749d9b7ebbc (iov_iter: fix copy_page_from_iter_atomic() if KMAP_LOCAL_FORCE_MAP), Hugh correctly noted that if KMAP_LOCAL_FORCE_MAP is enabled, we must limit ourselves to PAGE_SIZE bytes per call to kmap_local(). The same problem exists in memcpy_from_folio(), memcpy_to_folio(), folio_zero_tail(), folio_fill_tail() and memcpy_from_file_folio(), so add folio_test_partial_kmap() to do this more succinctly. Link: https://lkml.kernel.org/r/20250514170607.3000994-2-willy@infradead.org Fixes: 00cdf76012ab ("mm: add memcpy_from_file_folio()") Signed-off-by: Matthew Wilcox (Oracle) <willy(a)infradead.org> Cc: Al Viro <viro(a)zeniv.linux.org.uk> Cc: Hugh Dickins <hughd(a)google.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- include/linux/highmem.h | 10 +++++----- include/linux/page-flags.h | 7 +++++++ 2 files changed, 12 insertions(+), 5 deletions(-) --- a/include/linux/highmem.h~highmem-add-folio_test_partial_kmap +++ a/include/linux/highmem.h @@ -461,7 +461,7 @@ static inline void memcpy_from_folio(cha const char *from = kmap_local_folio(folio, offset); size_t chunk = len; - if (folio_test_highmem(folio) && + if (folio_test_partial_kmap(folio) && chunk > PAGE_SIZE - offset_in_page(offset)) chunk = PAGE_SIZE - offset_in_page(offset); memcpy(to, from, chunk); @@ -489,7 +489,7 @@ static inline void memcpy_to_folio(struc char *to = kmap_local_folio(folio, offset); size_t chunk = len; - if (folio_test_highmem(folio) && + if (folio_test_partial_kmap(folio) && chunk > PAGE_SIZE - offset_in_page(offset)) chunk = PAGE_SIZE - offset_in_page(offset); memcpy(to, from, chunk); @@ -522,7 +522,7 @@ static inline __must_check void *folio_z { size_t len = folio_size(folio) - offset; - if (folio_test_highmem(folio)) { + if (folio_test_partial_kmap(folio)) { size_t max = PAGE_SIZE - offset_in_page(offset); while (len > max) { @@ -560,7 +560,7 @@ static inline void folio_fill_tail(struc VM_BUG_ON(offset + len > folio_size(folio)); - if (folio_test_highmem(folio)) { + if (folio_test_partial_kmap(folio)) { size_t max = PAGE_SIZE - offset_in_page(offset); while (len > max) { @@ -597,7 +597,7 @@ static inline size_t memcpy_from_file_fo size_t offset = offset_in_folio(folio, pos); char *from = kmap_local_folio(folio, offset); - if (folio_test_highmem(folio)) { + if (folio_test_partial_kmap(folio)) { offset = offset_in_page(offset); len = min_t(size_t, len, PAGE_SIZE - offset); } else --- a/include/linux/page-flags.h~highmem-add-folio_test_partial_kmap +++ a/include/linux/page-flags.h @@ -615,6 +615,13 @@ FOLIO_FLAG(dropbehind, FOLIO_HEAD_PAGE) PAGEFLAG_FALSE(HighMem, highmem) #endif +/* Does kmap_local_folio() only allow access to one page of the folio? */ +#ifdef CONFIG_DEBUG_KMAP_LOCAL_FORCE_MAP +#define folio_test_partial_kmap(f) true +#else +#define folio_test_partial_kmap(f) folio_test_highmem(f) +#endif + #ifdef CONFIG_SWAP static __always_inline bool folio_test_swapcache(const struct folio *folio) { _ Patches currently in -mm which might be from willy(a)infradead.org are highmem-add-folio_test_partial_kmap.patch mm-rename-page-index-to-page-__folio_index.patch

3 months, 4 weeks

1
0
0 0

[PATCH] arm64: dts: qcom: sm8350: Fix typo in pil_camera_mem node

by Alok Tiwari

There is a typo in sm8350.dts where the node label mmeory@85200000 should be memory@85200000. This patch corrects the typo for clarity and consistency. Fixes: b7e8f433a673 ("arm64: dts: qcom: Add basic devicetree support for SM8350 SoC") Cc: stable(a)vger.kernel.org Signed-off-by: Alok Tiwari <alok.a.tiwari(a)oracle.com> --- arch/arm64/boot/dts/qcom/sm8350.dtsi | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/arm64/boot/dts/qcom/sm8350.dtsi b/arch/arm64/boot/dts/qcom/sm8350.dtsi index f2e12da13e68..971c828a7555 100644 --- a/arch/arm64/boot/dts/qcom/sm8350.dtsi +++ b/arch/arm64/boot/dts/qcom/sm8350.dtsi @@ -457,7 +457,7 @@ cdsp_secure_heap: memory@80c00000 { no-map; }; - pil_camera_mem: mmeory@85200000 { + pil_camera_mem: memory@85200000 { reg = <0x0 0x85200000 0x0 0x500000>; no-map; }; -- 2.47.1

3 months, 4 weeks

2
1
0 0

[PATCH 5.15 v2 00/14] ITS mitigation

by Pawan Gupta

v2: - Added the missing "--from" during patch generation. - Added missing patch to 6.1 backport. This is a backport of mitigation for Indirect Target Selection (ITS). ITS is a bug in some Intel CPUs that affects indirect branches including RETs in the first half of a cacheline. Mitigation is to relocate the affected branches to an ITS-safe thunk. Below additional upstream commits are required to cover some of the special cases like indirects in asm and returns in static calls: cfceff8526a4 ("x86/speculation: Simplify and make CALL_NOSPEC consistent") 052040e34c08 ("x86/speculation: Add a conditional CS prefix to CALL_NOSPEC") c8c81458863a ("x86/speculation: Remove the extra #ifdef around CALL_NOSPEC") d2408e043e72 ("x86/alternative: Optimize returns patching") 4ba89dd6ddec ("x86/alternatives: Remove faulty optimization") [1] https://github.com/torvalds/linux/commit/6f5bf947bab06f37ff931c359fd5770c4d… --- Borislav Petkov (AMD) (1): x86/alternative: Optimize returns patching Josh Poimboeuf (1): x86/alternatives: Remove faulty optimization Pawan Gupta (10): x86/speculation: Simplify and make CALL_NOSPEC consistent x86/speculation: Add a conditional CS prefix to CALL_NOSPEC x86/speculation: Remove the extra #ifdef around CALL_NOSPEC Documentation: x86/bugs/its: Add ITS documentation x86/its: Enumerate Indirect Target Selection (ITS) bug x86/its: Add support for ITS-safe indirect thunk x86/its: Add support for ITS-safe return thunk x86/its: Enable Indirect Target Selection mitigation x86/its: Add "vmexit" option to skip mitigation on some CPUs x86/its: Align RETs in BHB clear sequence to avoid thunking Peter Zijlstra (2): x86,nospec: Simplify {JMP,CALL}_NOSPEC x86/its: Use dynamic thunks for indirect branches Documentation/ABI/testing/sysfs-devices-system-cpu | 1 + Documentation/admin-guide/hw-vuln/index.rst | 1 + .../hw-vuln/indirect-target-selection.rst | 156 ++++++++++++++ Documentation/admin-guide/kernel-parameters.txt | 15 ++ arch/x86/Kconfig | 11 + arch/x86/entry/entry_64.S | 20 +- arch/x86/include/asm/alternative.h | 24 +++ arch/x86/include/asm/cpufeatures.h | 3 + arch/x86/include/asm/msr-index.h | 8 + arch/x86/include/asm/nospec-branch.h | 57 ++++-- arch/x86/kernel/alternative.c | 226 ++++++++++++++++++++- arch/x86/kernel/cpu/bugs.c | 139 ++++++++++++- arch/x86/kernel/cpu/common.c | 63 ++++-- arch/x86/kernel/ftrace.c | 2 +- arch/x86/kernel/module.c | 7 + arch/x86/kernel/static_call.c | 2 +- arch/x86/kernel/vmlinux.lds.S | 10 + arch/x86/kvm/x86.c | 4 +- arch/x86/lib/retpoline.S | 39 ++++ arch/x86/net/bpf_jit_comp.c | 8 +- drivers/base/cpu.c | 8 + include/linux/cpu.h | 2 + include/linux/module.h | 5 + 23 files changed, 768 insertions(+), 43 deletions(-) --- change-id: 20250512-its-5-15-0e0385221e32

3 months, 4 weeks

2
28
0 0

[PATCH 6.1 0/3] ITS fixes

by Pawan Gupta

Two of the patches are older rethunk fixes and one is a build fix for CONFIG_MODULES=n. --- Borislav Petkov (AMD) (1): x86/alternative: Optimize returns patching Eric Biggers (1): x86/its: Fix build errors when CONFIG_MODULES=n Josh Poimboeuf (1): x86/alternatives: Remove faulty optimization arch/x86/kernel/alternative.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) --- change-id: 20250513-its-fixes-6-1-d21ce20b0d1d

3 months, 4 weeks

2
6
0 0

[PATCH 6.1.y] bnxt_en: Fix receive ring space parameters when XDP is active

by Zhaoyang Li

From: Shravya KN <shravya.k-n(a)broadcom.com> [ Upstream commit 3051a77a09dfe3022aa012071346937fdf059033 ] The MTU setting at the time an XDP multi-buffer is attached determines whether the aggregation ring will be used and the rx_skb_func handler. This is done in bnxt_set_rx_skb_mode(). If the MTU is later changed, the aggregation ring setting may need to be changed and it may become out-of-sync with the settings initially done in bnxt_set_rx_skb_mode(). This may result in random memory corruption and crashes as the HW may DMA data larger than the allocated buffer size, such as: BUG: kernel NULL pointer dereference, address: 00000000000003c0 PGD 0 P4D 0 Oops: 0000 [#1] PREEMPT SMP NOPTI CPU: 17 PID: 0 Comm: swapper/17 Kdump: loaded Tainted: G S OE 6.1.0-226bf9805506 #1 Hardware name: Wiwynn Delta Lake PVT BZA.02601.0150/Delta Lake-Class1, BIOS F0E_3A12 08/26/2021 RIP: 0010:bnxt_rx_pkt+0xe97/0x1ae0 [bnxt_en] Code: 8b 95 70 ff ff ff 4c 8b 9d 48 ff ff ff 66 41 89 87 b4 00 00 00 e9 0b f7 ff ff 0f b7 43 0a 49 8b 95 a8 04 00 00 25 ff 0f 00 00 <0f> b7 14 42 48 c1 e2 06 49 03 95 a0 04 00 00 0f b6 42 33f RSP: 0018:ffffa19f40cc0d18 EFLAGS: 00010202 RAX: 00000000000001e0 RBX: ffff8e2c805c6100 RCX: 00000000000007ff RDX: 0000000000000000 RSI: ffff8e2c271ab990 RDI: ffff8e2c84f12380 RBP: ffffa19f40cc0e48 R08: 000000000001000d R09: 974ea2fcddfa4cbf R10: 0000000000000000 R11: ffffa19f40cc0ff8 R12: ffff8e2c94b58980 R13: ffff8e2c952d6600 R14: 0000000000000016 R15: ffff8e2c271ab990 FS: 0000000000000000(0000) GS:ffff8e3b3f840000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00000000000003c0 CR3: 0000000e8580a004 CR4: 00000000007706e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 PKRU: 55555554 Call Trace: <IRQ> __bnxt_poll_work+0x1c2/0x3e0 [bnxt_en] To address the issue, we now call bnxt_set_rx_skb_mode() within bnxt_change_mtu() to properly set the AGG rings configuration and update rx_skb_func based on the new MTU value. Additionally, BNXT_FLAG_NO_AGG_RINGS is cleared at the beginning of bnxt_set_rx_skb_mode() to make sure it gets set or cleared based on the current MTU. Fixes: 08450ea98ae9 ("bnxt_en: Fix max_mtu setting for multi-buf XDP") Co-developed-by: Somnath Kotur <somnath.kotur(a)broadcom.com> Signed-off-by: Somnath Kotur <somnath.kotur(a)broadcom.com> Signed-off-by: Shravya KN <shravya.k-n(a)broadcom.com> Signed-off-by: Michael Chan <michael.chan(a)broadcom.com> Signed-off-by: Paolo Abeni <pabeni(a)redhat.com> Signed-off-by: Zhaoyang Li <lizy04(a)hust.edu.cn> --- drivers/net/ethernet/broadcom/bnxt/bnxt.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.c b/drivers/net/ethernet/broadcom/bnxt/bnxt.c index 393a983f6d69..6b1245a3ab4b 100644 --- a/drivers/net/ethernet/broadcom/bnxt/bnxt.c +++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.c @@ -4041,7 +4041,7 @@ int bnxt_set_rx_skb_mode(struct bnxt *bp, bool page_mode) struct net_device *dev = bp->dev; if (page_mode) { - bp->flags &= ~BNXT_FLAG_AGG_RINGS; + bp->flags &= ~(BNXT_FLAG_AGG_RINGS | BNXT_FLAG_NO_AGG_RINGS); bp->flags |= BNXT_FLAG_RX_PAGE_MODE; if (bp->xdp_prog->aux->xdp_has_frags) @@ -12799,6 +12799,14 @@ static int bnxt_change_mtu(struct net_device *dev, int new_mtu) bnxt_close_nic(bp, true, false); dev->mtu = new_mtu; + + /* MTU change may change the AGG ring settings if an XDP multi-buffer + * program is attached. We need to set the AGG rings settings and + * rx_skb_func accordingly. + */ + if (READ_ONCE(bp->xdp_prog)) + bnxt_set_rx_skb_mode(bp, true); + bnxt_set_ring_params(bp); if (netif_running(dev)) -- 2.25.1

3 months, 4 weeks

2
1
0 0

FAILED: patch "[PATCH] drm/xe/gsc: do not flush the GSC worker from the reset path" failed to apply to 6.12-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 6.12-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. To reproduce the conflict and resubmit, you may use the following commands: git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-6.12.y git checkout FETCH_HEAD git cherry-pick -x 03552d8ac0afcc080c339faa0b726e2c0e9361cb # <resolve conflicts, build, test, etc.> git commit -s git send-email --to '<stable(a)vger.kernel.org>' --in-reply-to '2025051239-ceremony-rehab-8e6c@gregkh' --subject-prefix 'PATCH 6.12.y' HEAD^.. Possible dependencies: thanks, greg k-h ------------------ original commit in Linus's tree ------------------ From 03552d8ac0afcc080c339faa0b726e2c0e9361cb Mon Sep 17 00:00:00 2001 From: Daniele Ceraolo Spurio <daniele.ceraolospurio(a)intel.com> Date: Fri, 2 May 2025 08:51:04 -0700 Subject: [PATCH] drm/xe/gsc: do not flush the GSC worker from the reset path The workqueue used for the reset worker is marked as WQ_MEM_RECLAIM, while the GSC one isn't (and can't be as we need to do memory allocations in the gsc worker). Therefore, we can't flush the latter from the former. The reason why we had such a flush was to avoid interrupting either the GSC FW load or in progress GSC proxy operations. GSC proxy operations fall into 2 categories: 1) GSC proxy init: this only happens once immediately after GSC FW load and does not support being interrupted. The only way to recover from an interruption of the proxy init is to do an FLR and re-load the GSC. 2) GSC proxy request: this can happen in response to a request that the driver sends to the GSC. If this is interrupted, the GSC FW will timeout and the driver request will be failed, but overall the GSC will keep working fine. Flushing the work allowed us to avoid interruption in both cases (unless the hang came from the GSC engine itself, in which case we're toast anyway). However, a failure on a proxy request is tolerable if we're in a scenario where we're triggering a GT reset (i.e., something is already gone pretty wrong), so what we really need to avoid is interrupting the init flow, which we can do by polling on the register that reports when the proxy init is complete (as that ensure us that all the load and init operations have been completed). Note that during suspend we still want to do a flush of the worker to make sure it completes any operations involving the HW before the power is cut. v2: fix spelling in commit msg, rename waiter function (Julia) Fixes: dd0e89e5edc2 ("drm/xe/gsc: GSC FW load") Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/4830 Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio(a)intel.com> Cc: John Harrison <John.C.Harrison(a)Intel.com> Cc: Alan Previn <alan.previn.teres.alexis(a)intel.com> Cc: <stable(a)vger.kernel.org> # v6.8+ Reviewed-by: Julia Filipchuk <julia.filipchuk(a)intel.com> Link: https://lore.kernel.org/r/20250502155104.2201469-1-daniele.ceraolospurio@in… (cherry picked from commit 12370bfcc4f0bdf70279ec5b570eb298963422b5) Signed-off-by: Lucas De Marchi <lucas.demarchi(a)intel.com> diff --git a/drivers/gpu/drm/xe/xe_gsc.c b/drivers/gpu/drm/xe/xe_gsc.c index fd41113f8572..0bcf97063ff6 100644 --- a/drivers/gpu/drm/xe/xe_gsc.c +++ b/drivers/gpu/drm/xe/xe_gsc.c @@ -555,6 +555,28 @@ void xe_gsc_wait_for_worker_completion(struct xe_gsc *gsc) flush_work(&gsc->work); } +void xe_gsc_stop_prepare(struct xe_gsc *gsc) +{ + struct xe_gt *gt = gsc_to_gt(gsc); + int ret; + + if (!xe_uc_fw_is_loadable(&gsc->fw) || xe_uc_fw_is_in_error_state(&gsc->fw)) + return; + + xe_force_wake_assert_held(gt_to_fw(gt), XE_FW_GSC); + + /* + * If the GSC FW load or the proxy init are interrupted, the only way + * to recover it is to do an FLR and reload the GSC from scratch. + * Therefore, let's wait for the init to complete before stopping + * operations. The proxy init is the last step, so we can just wait on + * that + */ + ret = xe_gsc_wait_for_proxy_init_done(gsc); + if (ret) + xe_gt_err(gt, "failed to wait for GSC init completion before uc stop\n"); +} + /* * wa_14015076503: if the GSC FW is loaded, we need to alert it before doing a * GSC engine reset by writing a notification bit in the GS1 register and then diff --git a/drivers/gpu/drm/xe/xe_gsc.h b/drivers/gpu/drm/xe/xe_gsc.h index d99f66c38075..b8b8e0810ad9 100644 --- a/drivers/gpu/drm/xe/xe_gsc.h +++ b/drivers/gpu/drm/xe/xe_gsc.h @@ -16,6 +16,7 @@ struct xe_hw_engine; int xe_gsc_init(struct xe_gsc *gsc); int xe_gsc_init_post_hwconfig(struct xe_gsc *gsc); void xe_gsc_wait_for_worker_completion(struct xe_gsc *gsc); +void xe_gsc_stop_prepare(struct xe_gsc *gsc); void xe_gsc_load_start(struct xe_gsc *gsc); void xe_gsc_hwe_irq_handler(struct xe_hw_engine *hwe, u16 intr_vec); diff --git a/drivers/gpu/drm/xe/xe_gsc_proxy.c b/drivers/gpu/drm/xe/xe_gsc_proxy.c index 8cf70b228ff3..d0519cd6704a 100644 --- a/drivers/gpu/drm/xe/xe_gsc_proxy.c +++ b/drivers/gpu/drm/xe/xe_gsc_proxy.c @@ -71,6 +71,17 @@ bool xe_gsc_proxy_init_done(struct xe_gsc *gsc) HECI1_FWSTS1_PROXY_STATE_NORMAL; } +int xe_gsc_wait_for_proxy_init_done(struct xe_gsc *gsc) +{ + struct xe_gt *gt = gsc_to_gt(gsc); + + /* Proxy init can take up to 500ms, so wait double that for safety */ + return xe_mmio_wait32(&gt->mmio, HECI_FWSTS1(MTL_GSC_HECI1_BASE), + HECI1_FWSTS1_CURRENT_STATE, + HECI1_FWSTS1_PROXY_STATE_NORMAL, + USEC_PER_SEC, NULL, false); +} + static void __gsc_proxy_irq_rmw(struct xe_gsc *gsc, u32 clr, u32 set) { struct xe_gt *gt = gsc_to_gt(gsc); diff --git a/drivers/gpu/drm/xe/xe_gsc_proxy.h b/drivers/gpu/drm/xe/xe_gsc_proxy.h index fdef56995cd4..765602221dbc 100644 --- a/drivers/gpu/drm/xe/xe_gsc_proxy.h +++ b/drivers/gpu/drm/xe/xe_gsc_proxy.h @@ -12,6 +12,7 @@ struct xe_gsc; int xe_gsc_proxy_init(struct xe_gsc *gsc); bool xe_gsc_proxy_init_done(struct xe_gsc *gsc); +int xe_gsc_wait_for_proxy_init_done(struct xe_gsc *gsc); int xe_gsc_proxy_start(struct xe_gsc *gsc); int xe_gsc_proxy_request_handler(struct xe_gsc *gsc); diff --git a/drivers/gpu/drm/xe/xe_gt.c b/drivers/gpu/drm/xe/xe_gt.c index 10a9e3c72b36..66198cf2662c 100644 --- a/drivers/gpu/drm/xe/xe_gt.c +++ b/drivers/gpu/drm/xe/xe_gt.c @@ -857,7 +857,7 @@ void xe_gt_suspend_prepare(struct xe_gt *gt) fw_ref = xe_force_wake_get(gt_to_fw(gt), XE_FORCEWAKE_ALL); - xe_uc_stop_prepare(&gt->uc); + xe_uc_suspend_prepare(&gt->uc); xe_force_wake_put(gt_to_fw(gt), fw_ref); } diff --git a/drivers/gpu/drm/xe/xe_uc.c b/drivers/gpu/drm/xe/xe_uc.c index c14bd2282044..3a8751a8b92d 100644 --- a/drivers/gpu/drm/xe/xe_uc.c +++ b/drivers/gpu/drm/xe/xe_uc.c @@ -244,7 +244,7 @@ void xe_uc_gucrc_disable(struct xe_uc *uc) void xe_uc_stop_prepare(struct xe_uc *uc) { - xe_gsc_wait_for_worker_completion(&uc->gsc); + xe_gsc_stop_prepare(&uc->gsc); xe_guc_stop_prepare(&uc->guc); } @@ -278,6 +278,12 @@ static void uc_reset_wait(struct xe_uc *uc) goto again; } +void xe_uc_suspend_prepare(struct xe_uc *uc) +{ + xe_gsc_wait_for_worker_completion(&uc->gsc); + xe_guc_stop_prepare(&uc->guc); +} + int xe_uc_suspend(struct xe_uc *uc) { /* GuC submission not enabled, nothing to do */ diff --git a/drivers/gpu/drm/xe/xe_uc.h b/drivers/gpu/drm/xe/xe_uc.h index 3813c1ede450..c23e6f5e2514 100644 --- a/drivers/gpu/drm/xe/xe_uc.h +++ b/drivers/gpu/drm/xe/xe_uc.h @@ -18,6 +18,7 @@ int xe_uc_reset_prepare(struct xe_uc *uc); void xe_uc_stop_prepare(struct xe_uc *uc); void xe_uc_stop(struct xe_uc *uc); int xe_uc_start(struct xe_uc *uc); +void xe_uc_suspend_prepare(struct xe_uc *uc); int xe_uc_suspend(struct xe_uc *uc); int xe_uc_sanitize_reset(struct xe_uc *uc); void xe_uc_declare_wedged(struct xe_uc *uc);

3 months, 4 weeks

3
2
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror May 2025