emit_glsl_warp: Fix shuffle ops for 64-thread warp sizes

Project <tentative title>: Rework Garbage Collection.

src/common/logging/filter.cpp

@@ -111,6 +111,7 @@ bool ParseFilterRule(Filter& instance, Iterator begin, Iterator end) {
     SUB(Service, NCM)                                                                              \
     SUB(Service, NFC)                                                                              \
     SUB(Service, NFP)                                                                              \
+    SUB(Service, NGCT)                                                                             \
     SUB(Service, NIFM)                                                                             \
     SUB(Service, NIM)                                                                              \
     SUB(Service, NPNS)                                                                             \

src/common/logging/types.h

@@ -81,6 +81,7 @@ enum class Class : u8 {
     Service_NCM,       ///< The NCM service
     Service_NFC,       ///< The NFC (Near-field communication) service
     Service_NFP,       ///< The NFP service
+    Service_NGCT,      ///< The NGCT (No Good Content for Terra) service
     Service_NIFM,      ///< The NIFM (Network interface) service
     Service_NIM,       ///< The NIM service
     Service_NPNS,      ///< The NPNS service

src/core/CMakeLists.txt

@@ -452,6 +452,8 @@ add_library(core STATIC
     hle/service/nfp/nfp.h
     hle/service/nfp/nfp_user.cpp
     hle/service/nfp/nfp_user.h
+    hle/service/ngct/ngct.cpp
+    hle/service/ngct/ngct.h
     hle/service/nifm/nifm.cpp
     hle/service/nifm/nifm.h
     hle/service/nim/nim.cpp

src/core/hle/service/ngct/ngct.cpp

@@ -0,0 +1,46 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included
+
+#include "common/string_util.h"
+#include "core/core.h"
+#include "core/hle/ipc_helpers.h"
+#include "core/hle/service/ngct/ngct.h"
+#include "core/hle/service/service.h"
+
+namespace Service::NGCT {
+
+class IService final : public ServiceFramework<IService> {
+public:
+    explicit IService(Core::System& system_) : ServiceFramework{system_, "ngct:u"} {
+        // clang-format off
+        static const FunctionInfo functions[] = {
+            {0, nullptr, "Match"},
+            {1, &IService::Filter, "Filter"},
+        };
+        // clang-format on
+
+        RegisterHandlers(functions);
+    }
+
+private:
+    void Filter(Kernel::HLERequestContext& ctx) {
+        const auto buffer = ctx.ReadBuffer();
+        const auto text = Common::StringFromFixedZeroTerminatedBuffer(
+            reinterpret_cast<const char*>(buffer.data()), buffer.size());
+
+        LOG_WARNING(Service_NGCT, "(STUBBED) called, text={}", text);
+
+        // Return the same string since we don't censor anything
+        ctx.WriteBuffer(buffer);
+
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(ResultSuccess);
+    }
+};
+
+void InstallInterfaces(SM::ServiceManager& service_manager, Core::System& system) {
+    std::make_shared<IService>(system)->InstallAsService(system.ServiceManager());
+}
+
+} // namespace Service::NGCT

src/core/hle/service/ngct/ngct.h

@@ -0,0 +1,20 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included
+
+#pragma once
+
+namespace Core {
+class System;
+}
+
+namespace Service::SM {
+class ServiceManager;
+}
+
+namespace Service::NGCT {
+
+/// Registers all NGCT services with the specified service manager.
+void InstallInterfaces(SM::ServiceManager& service_manager, Core::System& system);
+
+} // namespace Service::NGCT

src/core/hle/service/nifm/nifm.cpp

@@ -277,37 +277,45 @@ private:
     void GetCurrentNetworkProfile(Kernel::HLERequestContext& ctx) {
         LOG_WARNING(Service_NIFM, "(STUBBED) called");
 
-        const SfNetworkProfileData network_profile_data{
-            .ip_setting_data{
-                .ip_address_setting{
-                    .is_automatic{true},
-                    .current_address{192, 168, 1, 100},
-                    .subnet_mask{255, 255, 255, 0},
-                    .gateway{192, 168, 1, 1},
+        const auto net_iface = Network::GetSelectedNetworkInterface();
+
+        const SfNetworkProfileData network_profile_data = [&net_iface] {
+            if (!net_iface) {
+                return SfNetworkProfileData{};
+            }
+
+            return SfNetworkProfileData{
+                .ip_setting_data{
+                    .ip_address_setting{
+                        .is_automatic{true},
+                        .current_address{Network::TranslateIPv4(net_iface->ip_address)},
+                        .subnet_mask{Network::TranslateIPv4(net_iface->subnet_mask)},
+                        .gateway{Network::TranslateIPv4(net_iface->gateway)},
+                    },
+                    .dns_setting{
+                        .is_automatic{true},
+                        .primary_dns{1, 1, 1, 1},
+                        .secondary_dns{1, 0, 0, 1},
+                    },
+                    .proxy_setting{
+                        .enabled{false},
+                        .port{},
+                        .proxy_server{},
+                        .automatic_auth_enabled{},
+                        .user{},
+                        .password{},
+                    },
+                    .mtu{1500},
                 },
-                .dns_setting{
-                    .is_automatic{true},
-                    .primary_dns{1, 1, 1, 1},
-                    .secondary_dns{1, 0, 0, 1},
+                .uuid{0xdeadbeef, 0xdeadbeef},
+                .network_name{"yuzu Network"},
+                .wireless_setting_data{
+                    .ssid_length{12},
+                    .ssid{"yuzu Network"},
+                    .passphrase{"yuzupassword"},
                 },
-                .proxy_setting{
-                    .enabled{false},
-                    .port{},
-                    .proxy_server{},
-                    .automatic_auth_enabled{},
-                    .user{},
-                    .password{},
-                },
-                .mtu{1500},
-            },
-            .uuid{0xdeadbeef, 0xdeadbeef},
-            .network_name{"yuzu Network"},
-            .wireless_setting_data{
-                .ssid_length{12},
-                .ssid{"yuzu Network"},
-                .passphrase{"yuzupassword"},
-            },
-        };
+            };
+        }();
 
         ctx.WriteBuffer(network_profile_data);
 
@@ -352,38 +360,33 @@ private:
         LOG_WARNING(Service_NIFM, "(STUBBED) called");
 
         struct IpConfigInfo {
-            IpAddressSetting ip_address_setting;
-            DnsSetting dns_setting;
+            IpAddressSetting ip_address_setting{};
+            DnsSetting dns_setting{};
         };
         static_assert(sizeof(IpConfigInfo) == sizeof(IpAddressSetting) + sizeof(DnsSetting),
                       "IpConfigInfo has incorrect size.");
 
-        IpConfigInfo ip_config_info{
-            .ip_address_setting{
-                .is_automatic{true},
-                .current_address{0, 0, 0, 0},
-                .subnet_mask{255, 255, 255, 0},
-                .gateway{192, 168, 1, 1},
-            },
-            .dns_setting{
-                .is_automatic{true},
-                .primary_dns{1, 1, 1, 1},
-                .secondary_dns{1, 0, 0, 1},
-            },
-        };
+        const auto net_iface = Network::GetSelectedNetworkInterface();
 
-        const auto iface = Network::GetSelectedNetworkInterface();
-        if (iface) {
-            ip_config_info.ip_address_setting =
-                IpAddressSetting{.is_automatic{true},
-                                 .current_address{Network::TranslateIPv4(iface->ip_address)},
-                                 .subnet_mask{Network::TranslateIPv4(iface->subnet_mask)},
-                                 .gateway{Network::TranslateIPv4(iface->gateway)}};
+        const IpConfigInfo ip_config_info = [&net_iface] {
+            if (!net_iface) {
+                return IpConfigInfo{};
+            }
 
-        } else {
-            LOG_ERROR(Service_NIFM,
-                      "Couldn't get host network configuration info, using default values");
-        }
+            return IpConfigInfo{
+                .ip_address_setting{
+                    .is_automatic{true},
+                    .current_address{Network::TranslateIPv4(net_iface->ip_address)},
+                    .subnet_mask{Network::TranslateIPv4(net_iface->subnet_mask)},
+                    .gateway{Network::TranslateIPv4(net_iface->gateway)},
+                },
+                .dns_setting{
+                    .is_automatic{true},
+                    .primary_dns{1, 1, 1, 1},
+                    .secondary_dns{1, 0, 0, 1},
+                },
+            };
+        }();
 
         IPC::ResponseBuilder rb{ctx, 2 + (sizeof(IpConfigInfo) + 3) / sizeof(u32)};
         rb.Push(ResultSuccess);

src/core/hle/service/service.cpp

@@ -46,6 +46,7 @@
 #include "core/hle/service/ncm/ncm.h"
 #include "core/hle/service/nfc/nfc.h"
 #include "core/hle/service/nfp/nfp.h"
+#include "core/hle/service/ngct/ngct.h"
 #include "core/hle/service/nifm/nifm.h"
 #include "core/hle/service/nim/nim.h"
 #include "core/hle/service/npns/npns.h"
@@ -271,6 +272,7 @@ Services::Services(std::shared_ptr<SM::ServiceManager>& sm, Core::System& system
     NCM::InstallInterfaces(*sm, system);
     NFC::InstallInterfaces(*sm, system);
     NFP::InstallInterfaces(*sm, system);
+    NGCT::InstallInterfaces(*sm, system);
     NIFM::InstallInterfaces(*sm, system);
     NIM::InstallInterfaces(*sm, system);
     NPNS::InstallInterfaces(*sm, system);

src/core/network/network_interface.cpp

@@ -37,73 +37,73 @@ std::vector<NetworkInterface> GetAvailableNetworkInterfaces() {
             AF_INET, GAA_FLAG_SKIP_MULTICAST | GAA_FLAG_SKIP_DNS_SERVER | GAA_FLAG_INCLUDE_GATEWAYS,
             nullptr, adapter_addresses.data(), &buf_size);
 
-        if (ret == ERROR_BUFFER_OVERFLOW) {
-            adapter_addresses.resize((buf_size / sizeof(IP_ADAPTER_ADDRESSES)) + 1);
-        } else {
+        if (ret != ERROR_BUFFER_OVERFLOW) {
             break;
         }
+
+        adapter_addresses.resize((buf_size / sizeof(IP_ADAPTER_ADDRESSES)) + 1);
     }
 
-    if (ret == NO_ERROR) {
-        std::vector<NetworkInterface> result;
-
-        for (auto current_address = adapter_addresses.data(); current_address != nullptr;
-             current_address = current_address->Next) {
-            if (current_address->FirstUnicastAddress == nullptr ||
-                current_address->FirstUnicastAddress->Address.lpSockaddr == nullptr) {
-                continue;
-            }
-
-            if (current_address->OperStatus != IfOperStatusUp) {
-                continue;
-            }
-
-            const auto ip_addr = Common::BitCast<struct sockaddr_in>(
-                                     *current_address->FirstUnicastAddress->Address.lpSockaddr)
-                                     .sin_addr;
-
-            ULONG mask = 0;
-            if (ConvertLengthToIpv4Mask(current_address->FirstUnicastAddress->OnLinkPrefixLength,
-                                        &mask) != NO_ERROR) {
-                LOG_ERROR(Network, "Failed to convert IPv4 prefix length to subnet mask");
-                continue;
-            }
-
-            struct in_addr gateway = {.S_un{.S_addr{0}}};
-            if (current_address->FirstGatewayAddress != nullptr &&
-                current_address->FirstGatewayAddress->Address.lpSockaddr != nullptr) {
-                gateway = Common::BitCast<struct sockaddr_in>(
-                              *current_address->FirstGatewayAddress->Address.lpSockaddr)
-                              .sin_addr;
-            }
-
-            result.push_back(NetworkInterface{
-                .name{Common::UTF16ToUTF8(std::wstring{current_address->FriendlyName})},
-                .ip_address{ip_addr},
-                .subnet_mask = in_addr{.S_un{.S_addr{mask}}},
-                .gateway = gateway});
-        }
-
-        return result;
-    } else {
+    if (ret != NO_ERROR) {
         LOG_ERROR(Network, "Failed to get network interfaces with GetAdaptersAddresses");
         return {};
     }
+
+    std::vector<NetworkInterface> result;
+
+    for (auto current_address = adapter_addresses.data(); current_address != nullptr;
+         current_address = current_address->Next) {
+        if (current_address->FirstUnicastAddress == nullptr ||
+            current_address->FirstUnicastAddress->Address.lpSockaddr == nullptr) {
+            continue;
+        }
+
+        if (current_address->OperStatus != IfOperStatusUp) {
+            continue;
+        }
+
+        const auto ip_addr = Common::BitCast<struct sockaddr_in>(
+                                 *current_address->FirstUnicastAddress->Address.lpSockaddr)
+                                 .sin_addr;
+
+        ULONG mask = 0;
+        if (ConvertLengthToIpv4Mask(current_address->FirstUnicastAddress->OnLinkPrefixLength,
+                                    &mask) != NO_ERROR) {
+            LOG_ERROR(Network, "Failed to convert IPv4 prefix length to subnet mask");
+            continue;
+        }
+
+        struct in_addr gateway = {.S_un{.S_addr{0}}};
+        if (current_address->FirstGatewayAddress != nullptr &&
+            current_address->FirstGatewayAddress->Address.lpSockaddr != nullptr) {
+            gateway = Common::BitCast<struct sockaddr_in>(
+                          *current_address->FirstGatewayAddress->Address.lpSockaddr)
+                          .sin_addr;
+        }
+
+        result.emplace_back(NetworkInterface{
+            .name{Common::UTF16ToUTF8(std::wstring{current_address->FriendlyName})},
+            .ip_address{ip_addr},
+            .subnet_mask = in_addr{.S_un{.S_addr{mask}}},
+            .gateway = gateway});
+    }
+
+    return result;
 }
 
 #else
 
 std::vector<NetworkInterface> GetAvailableNetworkInterfaces() {
-    std::vector<NetworkInterface> result;
-
     struct ifaddrs* ifaddr = nullptr;
 
     if (getifaddrs(&ifaddr) != 0) {
         LOG_ERROR(Network, "Failed to get network interfaces with getifaddrs: {}",
                   std::strerror(errno));
-        return result;
+        return {};
     }
 
+    std::vector<NetworkInterface> result;
+
     for (auto ifa = ifaddr; ifa != nullptr; ifa = ifa->ifa_next) {
         if (ifa->ifa_addr == nullptr || ifa->ifa_netmask == nullptr) {
             continue;
@@ -117,55 +117,62 @@ std::vector<NetworkInterface> GetAvailableNetworkInterfaces() {
             continue;
         }
 
-        std::uint32_t gateway{0};
+        u32 gateway{};
+
         std::ifstream file{"/proc/net/route"};
-        if (file.is_open()) {
-
-            // ignore header
-            file.ignore(std::numeric_limits<std::streamsize>::max(), '\n');
-
-            bool gateway_found = false;
-
-            for (std::string line; std::getline(file, line);) {
-                std::istringstream iss{line};
-
-                std::string iface_name{};
-                iss >> iface_name;
-                if (iface_name != ifa->ifa_name) {
-                    continue;
-                }
-
-                iss >> std::hex;
-
-                std::uint32_t dest{0};
-                iss >> dest;
-                if (dest != 0) {
-                    // not the default route
-                    continue;
-                }
-
-                iss >> gateway;
-
-                std::uint16_t flags{0};
-                iss >> flags;
-
-                // flag RTF_GATEWAY (defined in <linux/route.h>)
-                if ((flags & 0x2) == 0) {
-                    continue;
-                }
-
-                gateway_found = true;
-                break;
-            }
-
-            if (!gateway_found) {
-                gateway = 0;
-            }
-        } else {
+        if (!file.is_open()) {
             LOG_ERROR(Network, "Failed to open \"/proc/net/route\"");
+
+            result.emplace_back(NetworkInterface{
+                .name{ifa->ifa_name},
+                .ip_address{Common::BitCast<struct sockaddr_in>(*ifa->ifa_addr).sin_addr},
+                .subnet_mask{Common::BitCast<struct sockaddr_in>(*ifa->ifa_netmask).sin_addr},
+                .gateway{in_addr{.s_addr = gateway}}});
+            continue;
         }
 
-        result.push_back(NetworkInterface{
+        // ignore header
+        file.ignore(std::numeric_limits<std::streamsize>::max(), '\n');
+
+        bool gateway_found = false;
+
+        for (std::string line; std::getline(file, line);) {
+            std::istringstream iss{line};
+
+            std::string iface_name;
+            iss >> iface_name;
+            if (iface_name != ifa->ifa_name) {
+                continue;
+            }
+
+            iss >> std::hex;
+
+            u32 dest{};
+            iss >> dest;
+            if (dest != 0) {
+                // not the default route
+                continue;
+            }
+
+            iss >> gateway;
+
+            u16 flags{};
+            iss >> flags;
+
+            // flag RTF_GATEWAY (defined in <linux/route.h>)
+            if ((flags & 0x2) == 0) {
+                continue;
+            }
+
+            gateway_found = true;
+            break;
+        }
+
+        if (!gateway_found) {
+            gateway = 0;
+        }
+
+        result.emplace_back(NetworkInterface{
             .name{ifa->ifa_name},
             .ip_address{Common::BitCast<struct sockaddr_in>(*ifa->ifa_addr).sin_addr},
             .subnet_mask{Common::BitCast<struct sockaddr_in>(*ifa->ifa_netmask).sin_addr},
@@ -180,11 +187,11 @@ std::vector<NetworkInterface> GetAvailableNetworkInterfaces() {
 #endif
 
 std::optional<NetworkInterface> GetSelectedNetworkInterface() {
-    const std::string& selected_network_interface = Settings::values.network_interface.GetValue();
+    const auto& selected_network_interface = Settings::values.network_interface.GetValue();
     const auto network_interfaces = Network::GetAvailableNetworkInterfaces();
     if (network_interfaces.size() == 0) {
         LOG_ERROR(Network, "GetAvailableNetworkInterfaces returned no interfaces");
-        return {};
+        return std::nullopt;
     }
 
     const auto res =
@@ -192,12 +199,12 @@ std::optional<NetworkInterface> GetSelectedNetworkInterface() {
             return iface.name == selected_network_interface;
         });
 
-    if (res != network_interfaces.end()) {
-        return *res;
-    } else {
+    if (res == network_interfaces.end()) {
         LOG_ERROR(Network, "Couldn't find selected interface \"{}\"", selected_network_interface);
-        return {};
+        return std::nullopt;
     }
+
+    return *res;
 }
 
 } // namespace Network

src/shader_recompiler/backend/glsl/emit_glsl_warp.cpp

@@ -11,6 +11,8 @@
 
 namespace Shader::Backend::GLSL {
 namespace {
+constexpr char THREAD_ID[]{"gl_SubGroupInvocationARB"};
+
 void SetInBoundsFlag(EmitContext& ctx, IR::Inst& inst) {
     IR::Inst* const in_bounds{inst.GetAssociatedPseudoOperation(IR::Opcode::GetInBoundsFromOp)};
     if (!in_bounds) {
@@ -43,84 +45,100 @@ void UseShuffleNv(EmitContext& ctx, IR::Inst& inst, std::string_view shfl_op,
     ctx.AddU32("{}={}({},{},{},shfl_in_bounds);", inst, shfl_op, value, index, width);
     SetInBoundsFlag(ctx, inst);
 }
+
+std::string_view BallotIndex(EmitContext& ctx) {
+    if (!ctx.profile.warp_size_potentially_larger_than_guest) {
+        return ".x";
+    }
+    return "[gl_SubGroupInvocationARB>>5]";
+}
+
+std::string GetMask(EmitContext& ctx, std::string_view mask) {
+    const auto ballot_index{BallotIndex(ctx)};
+    return fmt::format("uint(uvec2({}){})", mask, ballot_index);
+}
 } // Anonymous namespace
 
 void EmitLaneId(EmitContext& ctx, IR::Inst& inst) {
-    ctx.AddU32("{}=gl_SubGroupInvocationARB&31u;", inst);
+    ctx.AddU32("{}={}&31u;", inst, THREAD_ID);
 }
 
 void EmitVoteAll(EmitContext& ctx, IR::Inst& inst, std::string_view pred) {
     if (!ctx.profile.warp_size_potentially_larger_than_guest) {
         ctx.AddU1("{}=allInvocationsEqualARB({});", inst, pred);
-    } else {
-        const auto active_mask{fmt::format("uvec2(ballotARB(true))[gl_SubGroupInvocationARB]")};
-        const auto ballot{fmt::format("uvec2(ballotARB({}))[gl_SubGroupInvocationARB]", pred)};
-        ctx.AddU1("{}=({}&{})=={};", inst, ballot, active_mask, active_mask);
+        return;
     }
+    const auto ballot_index{BallotIndex(ctx)};
+    const auto active_mask{fmt::format("uvec2(ballotARB(true)){}", ballot_index)};
+    const auto ballot{fmt::format("uvec2(ballotARB({})){}", pred, ballot_index)};
+    ctx.AddU1("{}=({}&{})=={};", inst, ballot, active_mask, active_mask);
 }
 
 void EmitVoteAny(EmitContext& ctx, IR::Inst& inst, std::string_view pred) {
     if (!ctx.profile.warp_size_potentially_larger_than_guest) {
         ctx.AddU1("{}=anyInvocationARB({});", inst, pred);
-    } else {
-        const auto active_mask{fmt::format("uvec2(ballotARB(true))[gl_SubGroupInvocationARB]")};
-        const auto ballot{fmt::format("uvec2(ballotARB({}))[gl_SubGroupInvocationARB]", pred)};
-        ctx.AddU1("{}=({}&{})!=0u;", inst, ballot, active_mask, active_mask);
+        return;
     }
+    const auto ballot_index{BallotIndex(ctx)};
+    const auto active_mask{fmt::format("uvec2(ballotARB(true)){}", ballot_index)};
+    const auto ballot{fmt::format("uvec2(ballotARB({})){}", pred, ballot_index)};
+    ctx.AddU1("{}=({}&{})!=0u;", inst, ballot, active_mask, active_mask);
 }
 
 void EmitVoteEqual(EmitContext& ctx, IR::Inst& inst, std::string_view pred) {
     if (!ctx.profile.warp_size_potentially_larger_than_guest) {
         ctx.AddU1("{}=allInvocationsEqualARB({});", inst, pred);
-    } else {
-        const auto active_mask{fmt::format("uvec2(ballotARB(true))[gl_SubGroupInvocationARB]")};
-        const auto ballot{fmt::format("uvec2(ballotARB({}))[gl_SubGroupInvocationARB]", pred)};
-        const auto value{fmt::format("({}^{})", ballot, active_mask)};
-        ctx.AddU1("{}=({}==0)||({}=={});", inst, value, value, active_mask);
+        return;
     }
+    const auto ballot_index{BallotIndex(ctx)};
+    const auto active_mask{fmt::format("uvec2(ballotARB(true)){}", ballot_index)};
+    const auto ballot{fmt::format("uvec2(ballotARB({})){}", pred, ballot_index)};
+    const auto value{fmt::format("({}^{})", ballot, active_mask)};
+    ctx.AddU1("{}=({}==0)||({}=={});", inst, value, value, active_mask);
 }
 
 void EmitSubgroupBallot(EmitContext& ctx, IR::Inst& inst, std::string_view pred) {
-    if (!ctx.profile.warp_size_potentially_larger_than_guest) {
-        ctx.AddU32("{}=uvec2(ballotARB({})).x;", inst, pred);
-    } else {
-        ctx.AddU32("{}=uvec2(ballotARB({}))[gl_SubGroupInvocationARB];", inst, pred);
-    }
+    const auto ballot_index{BallotIndex(ctx)};
+    ctx.AddU32("{}=uvec2(ballotARB({})){};", inst, pred, ballot_index);
 }
 
 void EmitSubgroupEqMask(EmitContext& ctx, IR::Inst& inst) {
-    ctx.AddU32("{}=uint(gl_SubGroupEqMaskARB.x);", inst);
+    ctx.AddU32("{}={};", inst, GetMask(ctx, "gl_SubGroupEqMaskARB"));
 }
 
 void EmitSubgroupLtMask(EmitContext& ctx, IR::Inst& inst) {
-    ctx.AddU32("{}=uint(gl_SubGroupLtMaskARB.x);", inst);
+    ctx.AddU32("{}={};", inst, GetMask(ctx, "gl_SubGroupLtMaskARB"));
 }
 
 void EmitSubgroupLeMask(EmitContext& ctx, IR::Inst& inst) {
-    ctx.AddU32("{}=uint(gl_SubGroupLeMaskARB.x);", inst);
+    ctx.AddU32("{}={};", inst, GetMask(ctx, "gl_SubGroupLeMaskARB"));
 }
 
 void EmitSubgroupGtMask(EmitContext& ctx, IR::Inst& inst) {
-    ctx.AddU32("{}=uint(gl_SubGroupGtMaskARB.x);", inst);
+    ctx.AddU32("{}={};", inst, GetMask(ctx, "gl_SubGroupGtMaskARB"));
 }
 
 void EmitSubgroupGeMask(EmitContext& ctx, IR::Inst& inst) {
-    ctx.AddU32("{}=uint(gl_SubGroupGeMaskARB.x);", inst);
+    ctx.AddU32("{}={};", inst, GetMask(ctx, "gl_SubGroupGeMaskARB"));
 }
 
 void EmitShuffleIndex(EmitContext& ctx, IR::Inst& inst, std::string_view value,
-                      std::string_view index, std::string_view clamp,
-                      std::string_view segmentation_mask) {
+                      std::string_view index, std::string_view clamp, std::string_view seg_mask) {
     if (ctx.profile.support_gl_warp_intrinsics) {
-        UseShuffleNv(ctx, inst, "shuffleNV", value, index, clamp, segmentation_mask);
+        UseShuffleNv(ctx, inst, "shuffleNV", value, index, clamp, seg_mask);
         return;
     }
-    const auto not_seg_mask{fmt::format("(~{})", segmentation_mask)};
-    const auto thread_id{"gl_SubGroupInvocationARB"};
-    const auto min_thread_id{ComputeMinThreadId(thread_id, segmentation_mask)};
-    const auto max_thread_id{ComputeMaxThreadId(min_thread_id, clamp, not_seg_mask)};
+    const bool big_warp{ctx.profile.warp_size_potentially_larger_than_guest};
+    const auto is_upper_partition{"int(gl_SubGroupInvocationARB)>=32"};
+    const auto upper_index{fmt::format("{}?{}+32:{}", is_upper_partition, index, index)};
+    const auto upper_clamp{fmt::format("{}?{}+32:{}", is_upper_partition, clamp, clamp)};
 
-    const auto lhs{fmt::format("({}&{})", index, not_seg_mask)};
+    const auto not_seg_mask{fmt::format("(~{})", seg_mask)};
+    const auto min_thread_id{ComputeMinThreadId(THREAD_ID, seg_mask)};
+    const auto max_thread_id{
+        ComputeMaxThreadId(min_thread_id, big_warp ? upper_clamp : clamp, not_seg_mask)};
+
+    const auto lhs{fmt::format("({}&{})", big_warp ? upper_index : index, not_seg_mask)};
     const auto src_thread_id{fmt::format("({})|({})", lhs, min_thread_id)};
     ctx.Add("shfl_in_bounds=int({})<=int({});", src_thread_id, max_thread_id);
     SetInBoundsFlag(ctx, inst);
@@ -128,29 +146,34 @@ void EmitShuffleIndex(EmitContext& ctx, IR::Inst& inst, std::string_view value,
 }
 
 void EmitShuffleUp(EmitContext& ctx, IR::Inst& inst, std::string_view value, std::string_view index,
-                   std::string_view clamp, std::string_view segmentation_mask) {
+                   std::string_view clamp, std::string_view seg_mask) {
     if (ctx.profile.support_gl_warp_intrinsics) {
-        UseShuffleNv(ctx, inst, "shuffleUpNV", value, index, clamp, segmentation_mask);
+        UseShuffleNv(ctx, inst, "shuffleUpNV", value, index, clamp, seg_mask);
         return;
     }
-    const auto thread_id{"gl_SubGroupInvocationARB"};
-    const auto max_thread_id{GetMaxThreadId(thread_id, clamp, segmentation_mask)};
-    const auto src_thread_id{fmt::format("({}-{})", thread_id, index)};
+    const bool big_warp{ctx.profile.warp_size_potentially_larger_than_guest};
+    const auto is_upper_partition{"int(gl_SubGroupInvocationARB)>=32"};
+    const auto upper_clamp{fmt::format("{}?{}+32:{}", is_upper_partition, clamp, clamp)};
+
+    const auto max_thread_id{GetMaxThreadId(THREAD_ID, big_warp ? upper_clamp : clamp, seg_mask)};
+    const auto src_thread_id{fmt::format("({}-{})", THREAD_ID, index)};
     ctx.Add("shfl_in_bounds=int({})>=int({});", src_thread_id, max_thread_id);
     SetInBoundsFlag(ctx, inst);
     ctx.AddU32("{}=shfl_in_bounds?readInvocationARB({},{}):{};", inst, value, src_thread_id, value);
 }
 
 void EmitShuffleDown(EmitContext& ctx, IR::Inst& inst, std::string_view value,
-                     std::string_view index, std::string_view clamp,
-                     std::string_view segmentation_mask) {
+                     std::string_view index, std::string_view clamp, std::string_view seg_mask) {
     if (ctx.profile.support_gl_warp_intrinsics) {
-        UseShuffleNv(ctx, inst, "shuffleDownNV", value, index, clamp, segmentation_mask);
+        UseShuffleNv(ctx, inst, "shuffleDownNV", value, index, clamp, seg_mask);
         return;
     }
-    const auto thread_id{"gl_SubGroupInvocationARB"};
-    const auto max_thread_id{GetMaxThreadId(thread_id, clamp, segmentation_mask)};
-    const auto src_thread_id{fmt::format("({}+{})", thread_id, index)};
+    const bool big_warp{ctx.profile.warp_size_potentially_larger_than_guest};
+    const auto is_upper_partition{"int(gl_SubGroupInvocationARB)>=32"};
+    const auto upper_clamp{fmt::format("{}?{}+32:{}", is_upper_partition, clamp, clamp)};
+
+    const auto max_thread_id{GetMaxThreadId(THREAD_ID, big_warp ? upper_clamp : clamp, seg_mask)};
+    const auto src_thread_id{fmt::format("({}+{})", THREAD_ID, index)};
     ctx.Add("shfl_in_bounds=int({})<=int({});", src_thread_id, max_thread_id);
     SetInBoundsFlag(ctx, inst);
     ctx.AddU32("{}=shfl_in_bounds?readInvocationARB({},{}):{};", inst, value, src_thread_id, value);
@@ -158,14 +181,17 @@ void EmitShuffleDown(EmitContext& ctx, IR::Inst& inst, std::string_view value,
 
 void EmitShuffleButterfly(EmitContext& ctx, IR::Inst& inst, std::string_view value,
                           std::string_view index, std::string_view clamp,
-                          std::string_view segmentation_mask) {
+                          std::string_view seg_mask) {
     if (ctx.profile.support_gl_warp_intrinsics) {
-        UseShuffleNv(ctx, inst, "shuffleXorNV", value, index, clamp, segmentation_mask);
+        UseShuffleNv(ctx, inst, "shuffleXorNV", value, index, clamp, seg_mask);
         return;
     }
-    const auto thread_id{"gl_SubGroupInvocationARB"};
-    const auto max_thread_id{GetMaxThreadId(thread_id, clamp, segmentation_mask)};
-    const auto src_thread_id{fmt::format("({}^{})", thread_id, index)};
+    const bool big_warp{ctx.profile.warp_size_potentially_larger_than_guest};
+    const auto is_upper_partition{"int(gl_SubGroupInvocationARB)>=32"};
+    const auto upper_clamp{fmt::format("{}?{}+32:{}", is_upper_partition, clamp, clamp)};
+
+    const auto max_thread_id{GetMaxThreadId(THREAD_ID, big_warp ? upper_clamp : clamp, seg_mask)};
+    const auto src_thread_id{fmt::format("({}^{})", THREAD_ID, index)};
     ctx.Add("shfl_in_bounds=int({})<=int({});", src_thread_id, max_thread_id);
     SetInBoundsFlag(ctx, inst);
     ctx.AddU32("{}=shfl_in_bounds?readInvocationARB({},{}):{};", inst, value, src_thread_id, value);

src/shader_recompiler/backend/spirv/emit_spirv_context_get_set.cpp

@@ -333,8 +333,9 @@ Id EmitGetAttribute(EmitContext& ctx, IR::Attribute attr, Id vertex) {
             return ctx.OpBitcast(ctx.F32[1], ctx.OpISub(ctx.U32[1], index, base));
         }
     case IR::Attribute::FrontFace:
-        return ctx.OpSelect(ctx.U32[1], ctx.OpLoad(ctx.U1, ctx.front_face),
-                            ctx.Const(std::numeric_limits<u32>::max()), ctx.u32_zero_value);
+        return ctx.OpSelect(ctx.F32[1], ctx.OpLoad(ctx.U1, ctx.front_face),
+                            ctx.OpBitcast(ctx.F32[1], ctx.Const(std::numeric_limits<u32>::max())),
+                            ctx.f32_zero_value);
     case IR::Attribute::PointSpriteS:
         return ctx.OpLoad(ctx.F32[1],
                           ctx.OpAccessChain(ctx.input_f32, ctx.point_coord, ctx.u32_zero_value));

src/shader_recompiler/backend/spirv/emit_spirv_warp.cpp

@@ -7,8 +7,13 @@
 
 namespace Shader::Backend::SPIRV {
 namespace {
+Id GetThreadId(EmitContext& ctx) {
+    return ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id);
+}
+
 Id WarpExtract(EmitContext& ctx, Id value) {
-    const Id local_index{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+    const Id thread_id{GetThreadId(ctx)};
+    const Id local_index{ctx.OpShiftRightArithmetic(ctx.U32[1], thread_id, ctx.Const(5U))};
     return ctx.OpVectorExtractDynamic(ctx.U32[1], value, local_index);
 }
 
@@ -48,10 +53,17 @@ Id SelectValue(EmitContext& ctx, Id in_range, Id value, Id src_thread_id) {
     return ctx.OpSelect(ctx.U32[1], in_range,
                         ctx.OpSubgroupReadInvocationKHR(ctx.U32[1], value, src_thread_id), value);
 }
+
+Id GetUpperClamp(EmitContext& ctx, Id invocation_id, Id clamp) {
+    const Id thirty_two{ctx.Const(32u)};
+    const Id is_upper_partition{ctx.OpSGreaterThanEqual(ctx.U1, invocation_id, thirty_two)};
+    const Id upper_clamp{ctx.OpIAdd(ctx.U32[1], thirty_two, clamp)};
+    return ctx.OpSelect(ctx.U32[1], is_upper_partition, upper_clamp, clamp);
+}
 } // Anonymous namespace
 
 Id EmitLaneId(EmitContext& ctx) {
-    const Id id{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+    const Id id{GetThreadId(ctx)};
     if (!ctx.profile.warp_size_potentially_larger_than_guest) {
         return id;
     }
@@ -123,7 +135,15 @@ Id EmitSubgroupGeMask(EmitContext& ctx) {
 Id EmitShuffleIndex(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clamp,
                     Id segmentation_mask) {
     const Id not_seg_mask{ctx.OpNot(ctx.U32[1], segmentation_mask)};
-    const Id thread_id{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+    const Id thread_id{GetThreadId(ctx)};
+    if (ctx.profile.warp_size_potentially_larger_than_guest) {
+        const Id thirty_two{ctx.Const(32u)};
+        const Id is_upper_partition{ctx.OpSGreaterThanEqual(ctx.U1, thread_id, thirty_two)};
+        const Id upper_index{ctx.OpIAdd(ctx.U32[1], thirty_two, index)};
+        const Id upper_clamp{ctx.OpIAdd(ctx.U32[1], thirty_two, clamp)};
+        index = ctx.OpSelect(ctx.U32[1], is_upper_partition, upper_index, index);
+        clamp = ctx.OpSelect(ctx.U32[1], is_upper_partition, upper_clamp, clamp);
+    }
     const Id min_thread_id{ComputeMinThreadId(ctx, thread_id, segmentation_mask)};
     const Id max_thread_id{ComputeMaxThreadId(ctx, min_thread_id, clamp, not_seg_mask)};
 
@@ -137,7 +157,10 @@ Id EmitShuffleIndex(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id cla
 
 Id EmitShuffleUp(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clamp,
                  Id segmentation_mask) {
-    const Id thread_id{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+    const Id thread_id{GetThreadId(ctx)};
+    if (ctx.profile.warp_size_potentially_larger_than_guest) {
+        clamp = GetUpperClamp(ctx, thread_id, clamp);
+    }
     const Id max_thread_id{GetMaxThreadId(ctx, thread_id, clamp, segmentation_mask)};
     const Id src_thread_id{ctx.OpISub(ctx.U32[1], thread_id, index)};
     const Id in_range{ctx.OpSGreaterThanEqual(ctx.U1, src_thread_id, max_thread_id)};
@@ -148,7 +171,10 @@ Id EmitShuffleUp(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clamp,
 
 Id EmitShuffleDown(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clamp,
                    Id segmentation_mask) {
-    const Id thread_id{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+    const Id thread_id{GetThreadId(ctx)};
+    if (ctx.profile.warp_size_potentially_larger_than_guest) {
+        clamp = GetUpperClamp(ctx, thread_id, clamp);
+    }
     const Id max_thread_id{GetMaxThreadId(ctx, thread_id, clamp, segmentation_mask)};
     const Id src_thread_id{ctx.OpIAdd(ctx.U32[1], thread_id, index)};
     const Id in_range{ctx.OpSLessThanEqual(ctx.U1, src_thread_id, max_thread_id)};
@@ -159,7 +185,10 @@ Id EmitShuffleDown(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clam
 
 Id EmitShuffleButterfly(EmitContext& ctx, IR::Inst* inst, Id value, Id index, Id clamp,
                         Id segmentation_mask) {
-    const Id thread_id{ctx.OpLoad(ctx.U32[1], ctx.subgroup_local_invocation_id)};
+    const Id thread_id{GetThreadId(ctx)};
+    if (ctx.profile.warp_size_potentially_larger_than_guest) {
+        clamp = GetUpperClamp(ctx, thread_id, clamp);
+    }
     const Id max_thread_id{GetMaxThreadId(ctx, thread_id, clamp, segmentation_mask)};
     const Id src_thread_id{ctx.OpBitwiseXor(ctx.U32[1], thread_id, index)};
     const Id in_range{ctx.OpSLessThanEqual(ctx.U1, src_thread_id, max_thread_id)};

src/video_core/command_classes/codecs/vp9.cpp

@@ -742,6 +742,7 @@ VpxBitStreamWriter VP9::ComposeUncompressedHeader() {
     uncomp_writer.WriteDeltaQ(current_frame_info.uv_dc_delta_q);
     uncomp_writer.WriteDeltaQ(current_frame_info.uv_ac_delta_q);
 
+    ASSERT(!current_frame_info.segment_enabled);
     uncomp_writer.WriteBit(false); // Segmentation enabled (TODO).
 
     const s32 min_tile_cols_log2 = CalcMinLog2TileCols(current_frame_info.frame_size.width);

src/video_core/command_classes/codecs/vp9_types.h

@@ -22,7 +22,7 @@ struct Vp9FrameDimensions {
 };
 static_assert(sizeof(Vp9FrameDimensions) == 0x8, "Vp9 Vp9FrameDimensions is an invalid size");
 
-enum FrameFlags : u32 {
+enum class FrameFlags : u32 {
     IsKeyFrame = 1 << 0,
     LastFrameIsKeyFrame = 1 << 1,
     FrameSizeChanged = 1 << 2,
@@ -30,6 +30,7 @@ enum FrameFlags : u32 {
     LastShowFrame = 1 << 4,
     IntraOnly = 1 << 5,
 };
+DECLARE_ENUM_FLAG_OPERATORS(FrameFlags)
 
 enum class TxSize {
     Tx4x4 = 0,   // 4x4 transform
@@ -92,44 +93,34 @@ struct Vp9EntropyProbs {
 static_assert(sizeof(Vp9EntropyProbs) == 0x7B4, "Vp9EntropyProbs is an invalid size");
 
 struct Vp9PictureInfo {
-    bool is_key_frame;
-    bool intra_only;
-    bool last_frame_was_key;
-    bool frame_size_changed;
-    bool error_resilient_mode;
-    bool last_frame_shown;
-    bool show_frame;
+    u32 bitstream_size;
+    std::array<u64, 4> frame_offsets;
     std::array<s8, 4> ref_frame_sign_bias;
     s32 base_q_index;
     s32 y_dc_delta_q;
     s32 uv_dc_delta_q;
     s32 uv_ac_delta_q;
-    bool lossless;
     s32 transform_mode;
-    bool allow_high_precision_mv;
     s32 interp_filter;
     s32 reference_mode;
-    s8 comp_fixed_ref;
-    std::array<s8, 2> comp_var_ref;
     s32 log2_tile_cols;
     s32 log2_tile_rows;
-    bool segment_enabled;
-    bool segment_map_update;
-    bool segment_map_temporal_update;
-    s32 segment_abs_delta;
-    std::array<u32, 8> segment_feature_enable;
-    std::array<std::array<s16, 4>, 8> segment_feature_data;
-    bool mode_ref_delta_enabled;
-    bool use_prev_in_find_mv_refs;
     std::array<s8, 4> ref_deltas;
     std::array<s8, 2> mode_deltas;
     Vp9EntropyProbs entropy;
     Vp9FrameDimensions frame_size;
     u8 first_level;
     u8 sharpness_level;
-    u32 bitstream_size;
-    std::array<u64, 4> frame_offsets;
-    std::array<bool, 4> refresh_frame;
+    bool is_key_frame;
+    bool intra_only;
+    bool last_frame_was_key;
+    bool error_resilient_mode;
+    bool last_frame_shown;
+    bool show_frame;
+    bool lossless;
+    bool allow_high_precision_mv;
+    bool segment_enabled;
+    bool mode_ref_delta_enabled;
 };
 
 struct Vp9FrameContainer {
@@ -145,7 +136,7 @@ struct PictureInfo {
     Vp9FrameDimensions golden_frame_size;  ///< 0x50
     Vp9FrameDimensions alt_frame_size;     ///< 0x58
     Vp9FrameDimensions current_frame_size; ///< 0x60
-    u32 vp9_flags;                         ///< 0x68
+    FrameFlags vp9_flags;                  ///< 0x68
     std::array<s8, 4> ref_frame_sign_bias; ///< 0x6C
     u8 first_level;                        ///< 0x70
     u8 sharpness_level;                    ///< 0x71
@@ -158,60 +149,43 @@ struct PictureInfo {
     u8 allow_high_precision_mv;            ///< 0x78
     u8 interp_filter;                      ///< 0x79
     u8 reference_mode;                     ///< 0x7A
-    s8 comp_fixed_ref;                     ///< 0x7B
-    std::array<s8, 2> comp_var_ref;        ///< 0x7C
+    INSERT_PADDING_BYTES_NOINIT(3);        ///< 0x7B
     u8 log2_tile_cols;                     ///< 0x7E
     u8 log2_tile_rows;                     ///< 0x7F
     Segmentation segmentation;             ///< 0x80
     LoopFilter loop_filter;                ///< 0xE4
-    INSERT_PADDING_BYTES_NOINIT(5);        ///< 0xEB
-    u32 surface_params;                    ///< 0xF0
-    INSERT_PADDING_WORDS_NOINIT(3);        ///< 0xF4
+    INSERT_PADDING_BYTES_NOINIT(21);       ///< 0xEB
 
     [[nodiscard]] Vp9PictureInfo Convert() const {
         return {
-            .is_key_frame = (vp9_flags & FrameFlags::IsKeyFrame) != 0,
-            .intra_only = (vp9_flags & FrameFlags::IntraOnly) != 0,
-            .last_frame_was_key = (vp9_flags & FrameFlags::LastFrameIsKeyFrame) != 0,
-            .frame_size_changed = (vp9_flags & FrameFlags::FrameSizeChanged) != 0,
-            .error_resilient_mode = (vp9_flags & FrameFlags::ErrorResilientMode) != 0,
-            .last_frame_shown = (vp9_flags & FrameFlags::LastShowFrame) != 0,
-            .show_frame = true,
+            .bitstream_size = bitstream_size,
+            .frame_offsets{},
             .ref_frame_sign_bias = ref_frame_sign_bias,
             .base_q_index = base_q_index,
             .y_dc_delta_q = y_dc_delta_q,
             .uv_dc_delta_q = uv_dc_delta_q,
             .uv_ac_delta_q = uv_ac_delta_q,
-            .lossless = lossless != 0,
             .transform_mode = tx_mode,
-            .allow_high_precision_mv = allow_high_precision_mv != 0,
             .interp_filter = interp_filter,
             .reference_mode = reference_mode,
-            .comp_fixed_ref = comp_fixed_ref,
-            .comp_var_ref = comp_var_ref,
             .log2_tile_cols = log2_tile_cols,
             .log2_tile_rows = log2_tile_rows,
-            .segment_enabled = segmentation.enabled != 0,
-            .segment_map_update = segmentation.update_map != 0,
-            .segment_map_temporal_update = segmentation.temporal_update != 0,
-            .segment_abs_delta = segmentation.abs_delta,
-            .segment_feature_enable = segmentation.feature_mask,
-            .segment_feature_data = segmentation.feature_data,
-            .mode_ref_delta_enabled = loop_filter.mode_ref_delta_enabled != 0,
-            .use_prev_in_find_mv_refs = !(vp9_flags == (FrameFlags::ErrorResilientMode)) &&
-                                        !(vp9_flags == (FrameFlags::FrameSizeChanged)) &&
-                                        !(vp9_flags == (FrameFlags::IntraOnly)) &&
-                                        (vp9_flags == (FrameFlags::LastShowFrame)) &&
-                                        !(vp9_flags == (FrameFlags::LastFrameIsKeyFrame)),
             .ref_deltas = loop_filter.ref_deltas,
             .mode_deltas = loop_filter.mode_deltas,
             .entropy{},
             .frame_size = current_frame_size,
             .first_level = first_level,
             .sharpness_level = sharpness_level,
-            .bitstream_size = bitstream_size,
-            .frame_offsets{},
-            .refresh_frame{},
+            .is_key_frame = True(vp9_flags & FrameFlags::IsKeyFrame),
+            .intra_only = True(vp9_flags & FrameFlags::IntraOnly),
+            .last_frame_was_key = True(vp9_flags & FrameFlags::LastFrameIsKeyFrame),
+            .error_resilient_mode = True(vp9_flags & FrameFlags::ErrorResilientMode),
+            .last_frame_shown = True(vp9_flags & FrameFlags::LastShowFrame),
+            .show_frame = true,
+            .lossless = lossless != 0,
+            .allow_high_precision_mv = allow_high_precision_mv != 0,
+            .segment_enabled = segmentation.enabled != 0,
+            .mode_ref_delta_enabled = loop_filter.mode_ref_delta_enabled != 0,
         };
     }
 };
@@ -316,7 +290,6 @@ ASSERT_POSITION(last_frame_size, 0x48);
 ASSERT_POSITION(first_level, 0x70);
 ASSERT_POSITION(segmentation, 0x80);
 ASSERT_POSITION(loop_filter, 0xE4);
-ASSERT_POSITION(surface_params, 0xF0);
 #undef ASSERT_POSITION
 
 #define ASSERT_POSITION(field_name, position)                                                      \

src/video_core/renderer_vulkan/vk_blit_screen.cpp

@@ -159,11 +159,13 @@ VkSemaphore VKBlitScreen::Draw(const Tegra::FramebufferConfig& framebuffer,
 
         const VAddr framebuffer_addr = framebuffer.address + framebuffer.offset;
         const u8* const host_ptr = cpu_memory.GetPointer(framebuffer_addr);
-        const size_t size_bytes = GetSizeInBytes(framebuffer);
 
         // TODO(Rodrigo): Read this from HLE
         constexpr u32 block_height_log2 = 4;
         const u32 bytes_per_pixel = GetBytesPerPixel(framebuffer);
+        const u64 size_bytes{Tegra::Texture::CalculateSize(true, bytes_per_pixel,
+                                                           framebuffer.stride, framebuffer.height,
+                                                           1, block_height_log2, 0)};
         Tegra::Texture::UnswizzleTexture(
             mapped_span.subspan(image_offset, size_bytes), std::span(host_ptr, size_bytes),
             bytes_per_pixel, framebuffer.width, framebuffer.height, 1, block_height_log2, 0);

src/video_core/textures/decoders.cpp

@@ -63,14 +63,6 @@ void SwizzleImpl(std::span<u8> output, std::span<const u8> input, u32 width, u32
                 const u32 unswizzled_offset =
                     slice * pitch * height + line * pitch + column * BYTES_PER_PIXEL;
 
-                if (const auto offset = (TO_LINEAR ? unswizzled_offset : swizzled_offset);
-                    offset >= input.size()) {
-                    // TODO(Rodrigo): This is an out of bounds access that should never happen. To
-                    // avoid crashing the emulator, break.
-                    ASSERT_MSG(false, "offset {} exceeds input size {}!", offset, input.size());
-                    break;
-                }
-
                 u8* const dst = &output[TO_LINEAR ? swizzled_offset : unswizzled_offset];
                 const u8* const src = &input[TO_LINEAR ? unswizzled_offset : swizzled_offset];