November 2024 - Linux-stable-mirror

[PATCH v6 1/8] KVM: SVM: Fix gctx page leak on invalid inputs

by Dionna Glaze

Ensure that snp gctx page allocation is adequately deallocated on failure during snp_launch_start. Fixes: 136d8bc931c8 ("KVM: SEV: Add KVM_SEV_SNP_LAUNCH_START command") CC: Sean Christopherson <seanjc(a)google.com> CC: Paolo Bonzini <pbonzini(a)redhat.com> CC: Thomas Gleixner <tglx(a)linutronix.de> CC: Ingo Molnar <mingo(a)redhat.com> CC: Borislav Petkov <bp(a)alien8.de> CC: Dave Hansen <dave.hansen(a)linux.intel.com> CC: Ashish Kalra <ashish.kalra(a)amd.com> CC: Tom Lendacky <thomas.lendacky(a)amd.com> CC: John Allen <john.allen(a)amd.com> CC: Herbert Xu <herbert(a)gondor.apana.org.au> CC: "David S. Miller" <davem(a)davemloft.net> CC: Michael Roth <michael.roth(a)amd.com> CC: Luis Chamberlain <mcgrof(a)kernel.org> CC: Russ Weight <russ.weight(a)linux.dev> CC: Danilo Krummrich <dakr(a)redhat.com> CC: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> CC: "Rafael J. Wysocki" <rafael(a)kernel.org> CC: Tianfei zhang <tianfei.zhang(a)intel.com> CC: Alexey Kardashevskiy <aik(a)amd.com> CC: stable(a)vger.kernel.org Signed-off-by: Dionna Glaze <dionnaglaze(a)google.com> Acked-by: Sean Christopherson <seanjc(a)google.com> --- arch/x86/kvm/svm/sev.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c index c6c8524859001..357906375ec59 100644 --- a/arch/x86/kvm/svm/sev.c +++ b/arch/x86/kvm/svm/sev.c @@ -2212,10 +2212,6 @@ static int snp_launch_start(struct kvm *kvm, struct kvm_sev_cmd *argp) if (sev->snp_context) return -EINVAL; - sev->snp_context = snp_context_create(kvm, argp); - if (!sev->snp_context) - return -ENOTTY; - if (params.flags) return -EINVAL; @@ -2230,6 +2226,10 @@ static int snp_launch_start(struct kvm *kvm, struct kvm_sev_cmd *argp) if (params.policy & SNP_POLICY_MASK_SINGLE_SOCKET) return -EINVAL; + sev->snp_context = snp_context_create(kvm, argp); + if (!sev->snp_context) + return -ENOTTY; + start.gctx_paddr = __psp_pa(sev->snp_context); start.policy = params.policy; memcpy(start.gosvw, params.gosvw, sizeof(params.gosvw)); -- 2.47.0.277.g8800431eea-goog

10 months, 1 week

1
0
0 0

[PATCH] arm64: dts: rockchip: Fix vdd_gpu voltage constraints on PinePhone Pro

by Dragan Simic

The regulator-{min,max}-microvolt values for the vdd_gpu regulator in the PinePhone Pro device dts file are too restrictive, which prevents the highest GPU OPP from being used, slowing the GPU down unnecessarily. Let's fix that by making the regulator-{min,max}-microvolt values less strict, using the voltage range that the Silergy SYR838 chip used for the vdd_gpu regulator is actually capable of producing. [1][2] This also eliminates the following error messages from the kernel log: core: _opp_supported_by_regulators: OPP minuV: 1100000 maxuV: 1150000, not supported by regulator panfrost ff9a0000.gpu: _opp_add: OPP not supported by regulators (800000000) These changes to the regulator-{min,max}-microvolt values make the PinePhone Pro device dts consistent with the dts files for other Rockchip RK3399-based boards and devices. It's possible to be more strict here, by specifying the regulator-{min,max}-microvolt values that don't go outside of what the GPU actually may use, as the consumer of the vdd_gpu regulator, but those changes are left for a later directory-wide regulator cleanup. [1] https://files.pine64.org/doc/PinePhonePro/PinephonePro-Schematic-V1.0-20211… [2] https://www.t-firefly.com/download/Firefly-RK3399/docs/Chip%20Specification… Fixes: 78a21c7d5952 ("arm64: dts: rockchip: Add initial support for Pine64 PinePhone Pro") Cc: stable(a)vger.kernel.org Signed-off-by: Dragan Simic <dsimic(a)manjaro.org> --- arch/arm64/boot/dts/rockchip/rk3399-pinephone-pro.dts | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/arm64/boot/dts/rockchip/rk3399-pinephone-pro.dts b/arch/arm64/boot/dts/rockchip/rk3399-pinephone-pro.dts index 1a44582a49fb..956d64f5b271 100644 --- a/arch/arm64/boot/dts/rockchip/rk3399-pinephone-pro.dts +++ b/arch/arm64/boot/dts/rockchip/rk3399-pinephone-pro.dts @@ -410,8 +410,8 @@ vdd_gpu: regulator@41 { pinctrl-names = "default"; pinctrl-0 = <&vsel2_pin>; regulator-name = "vdd_gpu"; - regulator-min-microvolt = <875000>; - regulator-max-microvolt = <975000>; + regulator-min-microvolt = <712500>; + regulator-max-microvolt = <1500000>; regulator-ramp-delay = <1000>; regulator-always-on; regulator-boot-on;

10 months, 1 week

4
9
0 0

[PATCH 5.10/5.15/6.1 0/5] x86/mm: backport fixes for CVE-2023-0597 and CVE-2023-3640

by Vasiliy Kovalev

This series addresses two security vulnerabilities (CVE-2023-0597 [1], CVE-2023-3640 [2]) in the x86 memory management subsystem, alongside prerequisite [3] patches necessary for stable integration. [PATCH 5.10/5.15/6.1 1/5] x86/kasan: Map shadow for percpu pages on demand Ensures KASAN shadow mapping on demand for per-CPU pages. [PATCH 5.10/5.15/6.1 2/5] x86/mm: Recompute physical address for every page of per-CPU CEA mapping Calculates accurate physical addresses across CPU entry areas. [PATCH 5.10/5.15/6.1 3/5] x86/mm: Populate KASAN shadow for entire per-CPU range of CPU entry area Populates KASAN shadow memory for debugging across CPU entry areas. [PATCH 5.10/5.15/6.1 4/5] x86/mm: Randomize per-cpu entry area Randomizes the per-CPU entry area to reduce the risk of information leakage due to predictable memory layouts, especially in systems without KASLR, as described in CVE-2023-0597 [1]. [PATCH 5.10/5.15/6.1 5/5] x86/mm: Do not shuffle CPU entry areas without KASLR Prevents CPU entry area shuffling when KASLR is disabled, mitigating information leakage risks, as stated in CVE-2023-3640 [2]. [1] https://nvd.nist.gov/vuln/detail/CVE-2023-0597 [2] https://nvd.nist.gov/vuln/detail/CVE-2023-3640 [3] https://patchwork.ozlabs.org/project/ubuntu-kernel/cover/20230903234603.859…

10 months, 1 week

1
5
0 0

[PATCH v3] usb: dwc3: gadget: Add missing check for single port RAM in TxFIFO resizing logic

by Selvarasu Ganesan

The existing implementation of the TxFIFO resizing logic only supports scenarios where more than one port RAM is used. However, there is a need to resize the TxFIFO in USB2.0-only mode where only a single port RAM is available. This commit introduces the necessary changes to support TxFIFO resizing in such scenarios by adding a missing check for single port RAM. This fix addresses certain platform configurations where the existing TxFIFO resizing logic does not work properly due to the absence of support for single port RAM. By adding this missing check, we ensure that the TxFIFO resizing logic works correctly in all scenarios, including those with a single port RAM. Fixes: 9f607a309fbe ("usb: dwc3: Resize TX FIFOs to meet EP bursting requirements") Cc: stable(a)vger.kernel.org # 6.12.x: fad16c82: usb: dwc3: gadget: Refine the logic for resizing Tx FIFOs Signed-off-by: Selvarasu Ganesan <selvarasu.g(a)samsung.com> --- Changes in v3: - Updated the $subject and commit message. - Added Fixes tag, and addressed some minor comments from reviewer . - Link to v2: https://lore.kernel.org/linux-usb/20241111142049.604-1-selvarasu.g@samsung.… Changes in v2: - Removed the code change that limits the number of FIFOs for bulk EP, as plan to address this issue in a separate patch. - Renamed the variable spram_type to is_single_port_ram for better understanding. - Link to v1: https://lore.kernel.org/lkml/20241107104040.502-1-selvarasu.g@samsung.com/ --- drivers/usb/dwc3/core.h | 4 +++ drivers/usb/dwc3/gadget.c | 54 +++++++++++++++++++++++++++++++++------ 2 files changed, 50 insertions(+), 8 deletions(-) diff --git a/drivers/usb/dwc3/core.h b/drivers/usb/dwc3/core.h index eaa55c0cf62f..8306b39e5c64 100644 --- a/drivers/usb/dwc3/core.h +++ b/drivers/usb/dwc3/core.h @@ -915,6 +915,7 @@ struct dwc3_hwparams { #define DWC3_MODE(n) ((n) & 0x7) /* HWPARAMS1 */ +#define DWC3_SPRAM_TYPE(n) (((n) >> 23) & 1) #define DWC3_NUM_INT(n) (((n) & (0x3f << 15)) >> 15) /* HWPARAMS3 */ @@ -925,6 +926,9 @@ struct dwc3_hwparams { #define DWC3_NUM_IN_EPS(p) (((p)->hwparams3 & \ (DWC3_NUM_IN_EPS_MASK)) >> 18) +/* HWPARAMS6 */ +#define DWC3_RAM0_DEPTH(n) (((n) & (0xffff0000)) >> 16) + /* HWPARAMS7 */ #define DWC3_RAM1_DEPTH(n) ((n) & 0xffff) diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c index 2fed2aa01407..6101e5467b08 100644 --- a/drivers/usb/dwc3/gadget.c +++ b/drivers/usb/dwc3/gadget.c @@ -687,6 +687,44 @@ static int dwc3_gadget_calc_tx_fifo_size(struct dwc3 *dwc, int mult) return fifo_size; } +/** + * dwc3_gadget_calc_ram_depth - calculates the ram depth for txfifo + * @dwc: pointer to the DWC3 context + */ +static int dwc3_gadget_calc_ram_depth(struct dwc3 *dwc) +{ + int ram_depth; + int fifo_0_start; + bool is_single_port_ram; + + /* Check supporting RAM type by HW */ + is_single_port_ram = DWC3_SPRAM_TYPE(dwc->hwparams.hwparams1); + + /* + * If a single port RAM is utilized, then allocate TxFIFOs from + * RAM0. otherwise, allocate them from RAM1. + */ + ram_depth = is_single_port_ram ? DWC3_RAM0_DEPTH(dwc->hwparams.hwparams6) : + DWC3_RAM1_DEPTH(dwc->hwparams.hwparams7); + + /* + * In a single port RAM configuration, the available RAM is shared + * between the RX and TX FIFOs. This means that the txfifo can begin + * at a non-zero address. + */ + if (is_single_port_ram) { + u32 reg; + + /* Check if TXFIFOs start at non-zero addr */ + reg = dwc3_readl(dwc->regs, DWC3_GTXFIFOSIZ(0)); + fifo_0_start = DWC3_GTXFIFOSIZ_TXFSTADDR(reg); + + ram_depth -= (fifo_0_start >> 16); + } + + return ram_depth; +} + /** * dwc3_gadget_clear_tx_fifos - Clears txfifo allocation * @dwc: pointer to the DWC3 context @@ -753,7 +791,7 @@ static int dwc3_gadget_resize_tx_fifos(struct dwc3_ep *dep) { struct dwc3 *dwc = dep->dwc; int fifo_0_start; - int ram1_depth; + int ram_depth; int fifo_size; int min_depth; int num_in_ep; @@ -773,7 +811,7 @@ static int dwc3_gadget_resize_tx_fifos(struct dwc3_ep *dep) if (dep->flags & DWC3_EP_TXFIFO_RESIZED) return 0; - ram1_depth = DWC3_RAM1_DEPTH(dwc->hwparams.hwparams7); + ram_depth = dwc3_gadget_calc_ram_depth(dwc); switch (dwc->gadget->speed) { case USB_SPEED_SUPER_PLUS: @@ -809,7 +847,7 @@ static int dwc3_gadget_resize_tx_fifos(struct dwc3_ep *dep) /* Reserve at least one FIFO for the number of IN EPs */ min_depth = num_in_ep * (fifo + 1); - remaining = ram1_depth - min_depth - dwc->last_fifo_depth; + remaining = ram_depth - min_depth - dwc->last_fifo_depth; remaining = max_t(int, 0, remaining); /* * We've already reserved 1 FIFO per EP, so check what we can fit in @@ -835,9 +873,9 @@ static int dwc3_gadget_resize_tx_fifos(struct dwc3_ep *dep) dwc->last_fifo_depth += DWC31_GTXFIFOSIZ_TXFDEP(fifo_size); /* Check fifo size allocation doesn't exceed available RAM size. */ - if (dwc->last_fifo_depth >= ram1_depth) { + if (dwc->last_fifo_depth >= ram_depth) { dev_err(dwc->dev, "Fifosize(%d) > RAM size(%d) %s depth:%d\n", - dwc->last_fifo_depth, ram1_depth, + dwc->last_fifo_depth, ram_depth, dep->endpoint.name, fifo_size); if (DWC3_IP_IS(DWC3)) fifo_size = DWC3_GTXFIFOSIZ_TXFDEP(fifo_size); @@ -3090,7 +3128,7 @@ static int dwc3_gadget_check_config(struct usb_gadget *g) struct dwc3 *dwc = gadget_to_dwc(g); struct usb_ep *ep; int fifo_size = 0; - int ram1_depth; + int ram_depth; int ep_num = 0; if (!dwc->do_fifo_resize) @@ -3113,8 +3151,8 @@ static int dwc3_gadget_check_config(struct usb_gadget *g) fifo_size += dwc->max_cfg_eps; /* Check if we can fit a single fifo per endpoint */ - ram1_depth = DWC3_RAM1_DEPTH(dwc->hwparams.hwparams7); - if (fifo_size > ram1_depth) + ram_depth = dwc3_gadget_calc_ram_depth(dwc); + if (fifo_size > ram_depth) return -ENOMEM; return 0; -- 2.17.1

10 months, 1 week

2
1
0 0

backport "udf: Allocate name buffer in directory iterator on heap" to 5.15

by Hauke Mehrtens

Hi, I am running into this compile error in 5.15.171 in OpenWrt on 32 bit systems. This problem was introduced with kernel 5.15.169. ``` fs/udf/namei.c: In function 'udf_rename': fs/udf/namei.c:878:1: error: the frame size of 1144 bytes is larger than 1024 bytes [-Werror=frame-larger-than=] 878 | } | ^ cc1: all warnings being treated as errors make[2]: *** [scripts/Makefile.build:289: fs/udf/namei.o] Error 1 make[1]: *** [scripts/Makefile.build:552: fs/udf] Error 2 ``` This is fixed by this upstream commit: commit 0aba4860b0d0216a1a300484ff536171894d49d8 Author: Jan Kara <jack(a)suse.cz> Date: Tue Dec 20 12:38:45 2022 +0100 udf: Allocate name buffer in directory iterator on heap Please backport this patch to 5.15 too. It was already backported to kernel 6.1. Hauke

10 months, 1 week

2
2
0 0

[PATCH 8/9] drm/amd/display: Remove PIPE_DTO_SRC_SEL programming from set_dtbclk_dto

by Hamza Mahfooz

From: Ovidiu Bunea <Ovidiu.Bunea(a)amd.com> There are cases where an OTG is remapped from driving a regular HDMI display to a DP/eDP display. There are also cases where DTBCLK needs to be enabled for HPO, but DTBCLK DTO programming may be done while OTG is still enabled which is dangerous as the PIPE_DTO_SRC_SEL programming may change the pixel clock generator source for a mapped and running OTG and cause it to hang. Remove the PIPE_DTO_SRC_SEL programming from this sequence since it is already done in program_pixel_clk(). Additionally, make sure that program_pixel_clk sets DTBCLK DTO as source for special HDMI cases. Cc: stable(a)vger.kernel.org # 6.11+ Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas(a)amd.com> Signed-off-by: Ovidiu Bunea <Ovidiu.Bunea(a)amd.com> Signed-off-by: Hamza Mahfooz <hamza.mahfooz(a)amd.com> --- .../drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c | 15 +++++++++------ 1 file changed, 9 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c index 838d72eaa87f..b363f5360818 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c @@ -1392,10 +1392,10 @@ static void dccg35_set_dtbclk_dto( /* The recommended programming sequence to enable DTBCLK DTO to generate * valid pixel HPO DPSTREAM ENCODER, specifies that DTO source select should - * be set only after DTO is enabled + * be set only after DTO is enabled. + * PIPEx_DTO_SRC_SEL should not be programmed during DTBCLK update since OTG may still be on, and the + * programming is handled in program_pix_clk() regardless, so it can be removed from here. */ - REG_UPDATE(OTG_PIXEL_RATE_CNTL[params->otg_inst], - PIPE_DTO_SRC_SEL[params->otg_inst], 2); } else { switch (params->otg_inst) { case 0: @@ -1412,9 +1412,12 @@ static void dccg35_set_dtbclk_dto( break; } - REG_UPDATE_2(OTG_PIXEL_RATE_CNTL[params->otg_inst], - DTBCLK_DTO_ENABLE[params->otg_inst], 0, - PIPE_DTO_SRC_SEL[params->otg_inst], params->is_hdmi ? 0 : 1); + /** + * PIPEx_DTO_SRC_SEL should not be programmed during DTBCLK update since OTG may still be on, and the + * programming is handled in program_pix_clk() regardless, so it can be removed from here. + */ + REG_UPDATE(OTG_PIXEL_RATE_CNTL[params->otg_inst], + DTBCLK_DTO_ENABLE[params->otg_inst], 0); REG_WRITE(DTBCLK_DTO_MODULO[params->otg_inst], 0); REG_WRITE(DTBCLK_DTO_PHASE[params->otg_inst], 0); -- 2.46.1

10 months, 1 week

1
0
0 0

[PATCH 6/9] drm/amd/display: Populate Power Profile In Case of Early Return

by Hamza Mahfooz

From: Austin Zheng <Austin.Zheng(a)amd.com> Early return possible if context has no clk_mgr. This will lead to an invalid power profile being returned which looks identical to a profile with the lowest power level. Add back logic that populated the power profile and overwrite the value if needed. Cc: stable(a)vger.kernel.org Fixes: fc8c959496fa ("drm/amd/display: Update Interface to Check UCLK DPM") Reviewed-by: Dillon Varone <dillon.varone(a)amd.com> Signed-off-by: Austin Zheng <Austin.Zheng(a)amd.com> Signed-off-by: Hamza Mahfooz <hamza.mahfooz(a)amd.com> --- drivers/gpu/drm/amd/display/dc/core/dc.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 0c1875d35a95..1dd26d5df6b9 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -6100,11 +6100,11 @@ struct dc_power_profile dc_get_power_profile_for_dc_state(const struct dc_state { struct dc_power_profile profile = { 0 }; - if (!context || !context->clk_mgr || !context->clk_mgr->ctx || !context->clk_mgr->ctx->dc) + profile.power_level = !context->bw_ctx.bw.dcn.clk.p_state_change_support; + if (!context->clk_mgr || !context->clk_mgr->ctx || !context->clk_mgr->ctx->dc) return profile; struct dc *dc = context->clk_mgr->ctx->dc; - if (dc->res_pool->funcs->get_power_profile) profile.power_level = dc->res_pool->funcs->get_power_profile(context); return profile; -- 2.46.1

10 months, 1 week

1
0
0 0

[PATCH 4/9] drm/amd/display: Enable Request rate limiter during C-State on dcn401

by Hamza Mahfooz

From: Dillon Varone <dillon.varone(a)amd.com> [WHY] When C-State entry is requested, the rate limiter will be disabled which can result in high contention in the DCHUB return path. [HOW] Enable the rate limiter during C-state requests to prevent contention. Cc: stable(a)vger.kernel.org # 6.11+ Reviewed-by: Alvin Lee <alvin.lee2(a)amd.com> Signed-off-by: Dillon Varone <dillon.varone(a)amd.com> Signed-off-by: Hamza Mahfooz <hamza.mahfooz(a)amd.com> --- .../src/dml2_core/dml2_core_dcn4_calcs.c | 6 +++++ .../display/dc/hubbub/dcn10/dcn10_hubbub.h | 8 ++++++- .../display/dc/hubbub/dcn20/dcn20_hubbub.h | 1 + .../display/dc/hubbub/dcn401/dcn401_hubbub.c | 24 +++++++++++++++++-- .../display/dc/hubbub/dcn401/dcn401_hubbub.h | 7 +++++- .../amd/display/dc/hwss/dcn401/dcn401_hwseq.c | 13 ++++++---- .../gpu/drm/amd/display/dc/inc/hw/dchubbub.h | 2 +- .../dc/resource/dcn401/dcn401_resource.h | 3 ++- 8 files changed, 53 insertions(+), 11 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c index 92e43a1e4dd4..601320b1be81 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c @@ -11,6 +11,7 @@ #define DML2_MAX_FMT_420_BUFFER_WIDTH 4096 #define DML_MAX_NUM_OF_SLICES_PER_DSC 4 +#define ALLOW_SDPIF_RATE_LIMIT_PRE_CSTATE const char *dml2_core_internal_bw_type_str(enum dml2_core_internal_bw_type bw_type) { @@ -3886,6 +3887,10 @@ static void CalculateSwathAndDETConfiguration(struct dml2_core_internal_scratch #endif *p->hw_debug5 = false; +#ifdef ALLOW_SDPIF_RATE_LIMIT_PRE_CSTATE + if (p->NumberOfActiveSurfaces > 1) + *p->hw_debug5 = true; +#else for (unsigned int k = 0; k < p->NumberOfActiveSurfaces; ++k) { if (!(p->mrq_present) && (!(*p->UnboundedRequestEnabled)) && (TotalActiveDPP == 1) && p->display_cfg->plane_descriptors[k].surface.dcc.enable @@ -3901,6 +3906,7 @@ static void CalculateSwathAndDETConfiguration(struct dml2_core_internal_scratch dml2_printf("DML::%s: k=%u hw_debug5 = %u\n", __func__, k, *p->hw_debug5); #endif } +#endif } static enum dml2_odm_mode DecideODMMode(unsigned int HActive, diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn10/dcn10_hubbub.h b/drivers/gpu/drm/amd/display/dc/hubbub/dcn10/dcn10_hubbub.h index 4bd1dda07719..9fbd45c7dfef 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn10/dcn10_hubbub.h +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn10/dcn10_hubbub.h @@ -200,6 +200,7 @@ struct dcn_hubbub_registers { uint32_t DCHUBBUB_ARB_FRAC_URG_BW_MALL_B; uint32_t DCHUBBUB_TIMEOUT_DETECTION_CTRL1; uint32_t DCHUBBUB_TIMEOUT_DETECTION_CTRL2; + uint32_t DCHUBBUB_CTRL_STATUS; }; #define HUBBUB_REG_FIELD_LIST_DCN32(type) \ @@ -320,7 +321,12 @@ struct dcn_hubbub_registers { type DCHUBBUB_TIMEOUT_REQ_STALL_THRESHOLD;\ type DCHUBBUB_TIMEOUT_PSTATE_STALL_THRESHOLD;\ type DCHUBBUB_TIMEOUT_DETECTION_EN;\ - type DCHUBBUB_TIMEOUT_TIMER_RESET + type DCHUBBUB_TIMEOUT_TIMER_RESET;\ + type ROB_UNDERFLOW_STATUS;\ + type ROB_OVERFLOW_STATUS;\ + type ROB_OVERFLOW_CLEAR;\ + type DCHUBBUB_HW_DEBUG;\ + type CSTATE_SWATH_CHK_GOOD_MODE #define HUBBUB_STUTTER_REG_FIELD_LIST(type) \ type DCHUBBUB_ARB_ALLOW_SR_ENTER_WATERMARK_A;\ diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn20/dcn20_hubbub.h b/drivers/gpu/drm/amd/display/dc/hubbub/dcn20/dcn20_hubbub.h index 036bb3e6c957..46d8f5c70750 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn20/dcn20_hubbub.h +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn20/dcn20_hubbub.h @@ -96,6 +96,7 @@ struct dcn20_hubbub { unsigned int det1_size; unsigned int det2_size; unsigned int det3_size; + bool allow_sdpif_rate_limit_when_cstate_req; }; void hubbub2_construct(struct dcn20_hubbub *hubbub, diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn401/dcn401_hubbub.c b/drivers/gpu/drm/amd/display/dc/hubbub/dcn401/dcn401_hubbub.c index 5d658e9bef64..92fab471b183 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn401/dcn401_hubbub.c +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn401/dcn401_hubbub.c @@ -1192,15 +1192,35 @@ static void dcn401_wait_for_det_update(struct hubbub *hubbub, int hubp_inst) } } -static void dcn401_program_timeout_thresholds(struct hubbub *hubbub, struct dml2_display_arb_regs *arb_regs) +static bool dcn401_program_arbiter(struct hubbub *hubbub, struct dml2_display_arb_regs *arb_regs, bool safe_to_lower) { struct dcn20_hubbub *hubbub2 = TO_DCN20_HUBBUB(hubbub); + bool wm_pending = false; + uint32_t temp; + /* request backpressure and outstanding return threshold (unused)*/ //REG_UPDATE(DCHUBBUB_TIMEOUT_DETECTION_CTRL1, DCHUBBUB_TIMEOUT_REQ_STALL_THRESHOLD, arb_regs->req_stall_threshold); /* P-State stall threshold */ REG_UPDATE(DCHUBBUB_TIMEOUT_DETECTION_CTRL2, DCHUBBUB_TIMEOUT_PSTATE_STALL_THRESHOLD, arb_regs->pstate_stall_threshold); + + if (safe_to_lower || arb_regs->allow_sdpif_rate_limit_when_cstate_req > hubbub2->allow_sdpif_rate_limit_when_cstate_req) { + hubbub2->allow_sdpif_rate_limit_when_cstate_req = arb_regs->allow_sdpif_rate_limit_when_cstate_req; + + /* only update the required bits */ + REG_GET(DCHUBBUB_CTRL_STATUS, DCHUBBUB_HW_DEBUG, &temp); + if (hubbub2->allow_sdpif_rate_limit_when_cstate_req) { + temp |= (1 << 5); + } else { + temp &= ~(1 << 5); + } + REG_UPDATE(DCHUBBUB_CTRL_STATUS, DCHUBBUB_HW_DEBUG, temp); + } else { + wm_pending = true; + } + + return wm_pending; } static const struct hubbub_funcs hubbub4_01_funcs = { @@ -1226,7 +1246,7 @@ static const struct hubbub_funcs hubbub4_01_funcs = { .program_det_segments = dcn401_program_det_segments, .program_compbuf_segments = dcn401_program_compbuf_segments, .wait_for_det_update = dcn401_wait_for_det_update, - .program_timeout_thresholds = dcn401_program_timeout_thresholds, + .program_arbiter = dcn401_program_arbiter, }; void hubbub401_construct(struct dcn20_hubbub *hubbub2, diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn401/dcn401_hubbub.h b/drivers/gpu/drm/amd/display/dc/hubbub/dcn401/dcn401_hubbub.h index 5f1960722ebd..b1d9ea9d1c3d 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn401/dcn401_hubbub.h +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn401/dcn401_hubbub.h @@ -128,7 +128,12 @@ HUBBUB_SF(DCHUBBUB_TIMEOUT_DETECTION_CTRL1, DCHUBBUB_TIMEOUT_REQ_STALL_THRESHOLD, mask_sh),\ HUBBUB_SF(DCHUBBUB_TIMEOUT_DETECTION_CTRL2, DCHUBBUB_TIMEOUT_PSTATE_STALL_THRESHOLD, mask_sh),\ HUBBUB_SF(DCHUBBUB_TIMEOUT_DETECTION_CTRL2, DCHUBBUB_TIMEOUT_DETECTION_EN, mask_sh),\ - HUBBUB_SF(DCHUBBUB_TIMEOUT_DETECTION_CTRL2, DCHUBBUB_TIMEOUT_TIMER_RESET, mask_sh) + HUBBUB_SF(DCHUBBUB_TIMEOUT_DETECTION_CTRL2, DCHUBBUB_TIMEOUT_TIMER_RESET, mask_sh),\ + HUBBUB_SF(DCHUBBUB_CTRL_STATUS, ROB_UNDERFLOW_STATUS, mask_sh),\ + HUBBUB_SF(DCHUBBUB_CTRL_STATUS, ROB_OVERFLOW_STATUS, mask_sh),\ + HUBBUB_SF(DCHUBBUB_CTRL_STATUS, ROB_OVERFLOW_CLEAR, mask_sh),\ + HUBBUB_SF(DCHUBBUB_CTRL_STATUS, DCHUBBUB_HW_DEBUG, mask_sh),\ + HUBBUB_SF(DCHUBBUB_CTRL_STATUS, CSTATE_SWATH_CHK_GOOD_MODE, mask_sh) bool hubbub401_program_urgent_watermarks( struct hubbub *hubbub, diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c index e8cc1bfa73f3..5de11e2837c0 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c @@ -1488,6 +1488,10 @@ void dcn401_prepare_bandwidth(struct dc *dc, &context->bw_ctx.bw.dcn.watermarks, dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000, false); + /* update timeout thresholds */ + if (hubbub->funcs->program_arbiter) { + dc->wm_optimized_required |= hubbub->funcs->program_arbiter(hubbub, &context->bw_ctx.bw.dcn.arb_regs, false); + } /* decrease compbuf size */ if (hubbub->funcs->program_compbuf_segments) { @@ -1529,6 +1533,10 @@ void dcn401_optimize_bandwidth( &context->bw_ctx.bw.dcn.watermarks, dc->res_pool->ref_clocks.dchub_ref_clock_inKhz / 1000, true); + /* update timeout thresholds */ + if (hubbub->funcs->program_arbiter) { + hubbub->funcs->program_arbiter(hubbub, &context->bw_ctx.bw.dcn.arb_regs, true); + } if (dc->clk_mgr->dc_mode_softmax_enabled) if (dc->clk_mgr->clks.dramclk_khz > dc->clk_mgr->bw_params->dc_mode_softmax_memclk * 1000 && @@ -1554,11 +1562,6 @@ void dcn401_optimize_bandwidth( pipe_ctx->dlg_regs.min_dst_y_next_start); } } - - /* update timeout thresholds */ - if (hubbub->funcs->program_timeout_thresholds) { - hubbub->funcs->program_timeout_thresholds(hubbub, &context->bw_ctx.bw.dcn.arb_regs); - } } void dcn401_fams2_global_control_lock(struct dc *dc, diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h b/drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h index 6c1d41c0f099..52b745667ef7 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h @@ -228,7 +228,7 @@ struct hubbub_funcs { void (*program_det_segments)(struct hubbub *hubbub, int hubp_inst, unsigned det_buffer_size_seg); void (*program_compbuf_segments)(struct hubbub *hubbub, unsigned compbuf_size_seg, bool safe_to_increase); void (*wait_for_det_update)(struct hubbub *hubbub, int hubp_inst); - void (*program_timeout_thresholds)(struct hubbub *hubbub, struct dml2_display_arb_regs *arb_regs); + bool (*program_arbiter)(struct hubbub *hubbub, struct dml2_display_arb_regs *arb_regs, bool safe_to_lower); }; struct hubbub { diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h index 7c8d61db153d..19568c359669 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h @@ -612,7 +612,8 @@ void dcn401_prepare_mcache_programming(struct dc *dc, struct dc_state *context); SR(DCHUBBUB_SDPIF_CFG1), \ SR(DCHUBBUB_MEM_PWR_MODE_CTRL), \ SR(DCHUBBUB_TIMEOUT_DETECTION_CTRL1), \ - SR(DCHUBBUB_TIMEOUT_DETECTION_CTRL2) + SR(DCHUBBUB_TIMEOUT_DETECTION_CTRL2), \ + SR(DCHUBBUB_CTRL_STATUS) /* DCCG */ -- 2.46.1

10 months, 1 week

1
0
0 0

[PATCH 3/9] drm/amd/display: Fix handling of plane refcount

by Hamza Mahfooz

From: Joshua Aberback <joshua.aberback(a)amd.com> [Why] The mechanism to backup and restore plane states doesn't maintain refcount, which can cause issues if the refcount of the plane changes in between backup and restore operations, such as memory leaks if the refcount was supposed to go down, or double frees / invalid memory accesses if the refcount was supposed to go up. [How] Cache and re-apply current refcount when restoring plane states. Cc: stable(a)vger.kernel.org Reviewed-by: Josip Pavic <josip.pavic(a)amd.com> Signed-off-by: Joshua Aberback <joshua.aberback(a)amd.com> Signed-off-by: Hamza Mahfooz <hamza.mahfooz(a)amd.com> --- drivers/gpu/drm/amd/display/dc/core/dc.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 7872c6cabb14..0c1875d35a95 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -3141,7 +3141,10 @@ static void restore_planes_and_stream_state( return; for (i = 0; i < status->plane_count; i++) { + /* refcount will always be valid, restore everything else */ + struct kref refcount = status->plane_states[i]->refcount; *status->plane_states[i] = scratch->plane_states[i]; + status->plane_states[i]->refcount = refcount; } *stream = scratch->stream_state; } -- 2.46.1

10 months, 1 week

1
0
0 0

[PATCH 2/9] drm/amd/display: Ignore scalar validation failure if pipe is phantom

by Hamza Mahfooz

From: Chris Park <chris.park(a)amd.com> [Why] There are some pipe scaler validation failure when the pipe is phantom and causes crash in DML validation. Since, scalar parameters are not as important in phantom pipe and we require this plane to do successful MCLK switches, the failure condition can be ignored. [How] Ignore scalar validation failure if the pipe validation is marked as phantom pipe. Cc: stable(a)vger.kernel.org # 6.11+ Reviewed-by: Dillon Varone <dillon.varone(a)amd.com> Signed-off-by: Chris Park <chris.park(a)amd.com> Signed-off-by: Hamza Mahfooz <hamza.mahfooz(a)amd.com> --- drivers/gpu/drm/amd/display/dc/core/dc_resource.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c index 33125b95c3a1..619fad17de55 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c @@ -1501,6 +1501,10 @@ bool resource_build_scaling_params(struct pipe_ctx *pipe_ctx) res = spl_calculate_scaler_params(spl_in, spl_out); // Convert respective out params from SPL to scaler data translate_SPL_out_params_to_pipe_ctx(pipe_ctx, spl_out); + + /* Ignore scaler failure if pipe context plane is phantom plane */ + if (!res && plane_state->is_phantom) + res = true; } else { #endif /* depends on h_active */ @@ -1571,6 +1575,10 @@ bool resource_build_scaling_params(struct pipe_ctx *pipe_ctx) &plane_state->scaling_quality); } + /* Ignore scaler failure if pipe context plane is phantom plane */ + if (!res && plane_state->is_phantom) + res = true; + if (res && (pipe_ctx->plane_res.scl_data.taps.v_taps != temp.v_taps || pipe_ctx->plane_res.scl_data.taps.h_taps != temp.h_taps || pipe_ctx->plane_res.scl_data.taps.v_taps_c != temp.v_taps_c || -- 2.46.1

10 months, 1 week

1
0
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror November 2024