From 15e8156713cc38031642fafc8baf7d53f19f2e83 Mon Sep 17 00:00:00 2001 From: Chen Ridong Date: Fri, 25 Oct 2024 06:09:42 +0000 Subject: [PATCH 01/16] mm: shrinker: avoid memleak in alloc_shrinker_info A memleak was found as below: unreferenced object 0xffff8881010d2a80 (size 32): comm "mkdir", pid 1559, jiffies 4294932666 hex dump (first 32 bytes): 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ 40 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 @............... backtrace (crc 2e7ef6fa): [] __kmalloc_node_noprof+0x394/0x470 [] alloc_shrinker_info+0x7b/0x1a0 [] mem_cgroup_css_online+0x11a/0x3b0 [] online_css+0x29/0xa0 [] cgroup_apply_control_enable+0x20d/0x360 [] cgroup_mkdir+0x168/0x5f0 [] kernfs_iop_mkdir+0x5e/0x90 [] vfs_mkdir+0x144/0x220 [] do_mkdirat+0x87/0x130 [] __x64_sys_mkdir+0x49/0x70 [] do_syscall_64+0x68/0x140 [] entry_SYSCALL_64_after_hwframe+0x76/0x7e alloc_shrinker_info(), when shrinker_unit_alloc() returns an errer, the info won't be freed. Just fix it. Link: https://lkml.kernel.org/r/20241025060942.1049263-1-chenridong@huaweicloud.com Fixes: 307bececcd12 ("mm: shrinker: add a secondary array for shrinker_info::{map, nr_deferred}") Signed-off-by: Chen Ridong Acked-by: Qi Zheng Acked-by: Roman Gushchin Acked-by: Vlastimil Babka Acked-by: Kirill A. Shutemov Reviewed-by: Dave Chinner Cc: Anshuman Khandual Cc: Muchun Song Cc: Wang Weiyang Cc: Signed-off-by: Andrew Morton --- mm/shrinker.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/mm/shrinker.c b/mm/shrinker.c index dc5d2a6fcfc4..4a93fd433689 100644 --- a/mm/shrinker.c +++ b/mm/shrinker.c @@ -76,19 +76,21 @@ void free_shrinker_info(struct mem_cgroup *memcg) int alloc_shrinker_info(struct mem_cgroup *memcg) { - struct shrinker_info *info; int nid, ret = 0; int array_size = 0; mutex_lock(&shrinker_mutex); array_size = shrinker_unit_size(shrinker_nr_max); for_each_node(nid) { - info = kvzalloc_node(sizeof(*info) + array_size, GFP_KERNEL, nid); + struct shrinker_info *info = kvzalloc_node(sizeof(*info) + array_size, + GFP_KERNEL, nid); if (!info) goto err; info->map_nr_max = shrinker_nr_max; - if (shrinker_unit_alloc(info, NULL, nid)) + if (shrinker_unit_alloc(info, NULL, nid)) { + kvfree(info); goto err; + } rcu_assign_pointer(memcg->nodeinfo[nid]->shrinker_info, info); } mutex_unlock(&shrinker_mutex); -- 2.51.0 From d4148aeab412432bf928f311eca8a2ba52bb05df Mon Sep 17 00:00:00 2001 From: Vlastimil Babka Date: Thu, 24 Oct 2024 17:12:29 +0200 Subject: [PATCH 02/16] mm, mmap: limit THP alignment of anonymous mappings to PMD-aligned sizes Since commit efa7df3e3bb5 ("mm: align larger anonymous mappings on THP boundaries") a mmap() of anonymous memory without a specific address hint and of at least PMD_SIZE will be aligned to PMD so that it can benefit from a THP backing page. However this change has been shown to regress some workloads significantly. [1] reports regressions in various spec benchmarks, with up to 600% slowdown of the cactusBSSN benchmark on some platforms. The benchmark seems to create many mappings of 4632kB, which would have merged to a large THP-backed area before commit efa7df3e3bb5 and now they are fragmented to multiple areas each aligned to PMD boundary with gaps between. The regression then seems to be caused mainly due to the benchmark's memory access pattern suffering from TLB or cache aliasing due to the aligned boundaries of the individual areas. Another known regression bisected to commit efa7df3e3bb5 is darktable [2] [3] and early testing suggests this patch fixes the regression there as well. To fix the regression but still try to benefit from THP-friendly anonymous mapping alignment, add a condition that the size of the mapping must be a multiple of PMD size instead of at least PMD size. In case of many odd-sized mapping like the cactusBSSN creates, those will stop being aligned and with gaps between, and instead naturally merge again. Link: https://lkml.kernel.org/r/20241024151228.101841-2-vbabka@suse.cz Fixes: efa7df3e3bb5 ("mm: align larger anonymous mappings on THP boundaries") Signed-off-by: Vlastimil Babka Reported-by: Michael Matz Debugged-by: Gabriel Krisman Bertazi Closes: https://bugzilla.suse.com/show_bug.cgi?id=1229012 [1] Reported-by: Matthias Bodenbinder Closes: https://bugzilla.kernel.org/show_bug.cgi?id=219366 [2] Closes: https://lore.kernel.org/all/2050f0d4-57b0-481d-bab8-05e8d48fed0c@leemhuis.info/ [3] Reviewed-by: Lorenzo Stoakes Reviewed-by: Yang Shi Cc: Rik van Riel Cc: Jann Horn Cc: Liam R. Howlett Cc: Petr Tesarik Cc: Thorsten Leemhuis Cc: Signed-off-by: Andrew Morton --- mm/mmap.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/mm/mmap.c b/mm/mmap.c index 1e0e34cb993f..9841b41e3c76 100644 --- a/mm/mmap.c +++ b/mm/mmap.c @@ -900,7 +900,8 @@ __get_unmapped_area(struct file *file, unsigned long addr, unsigned long len, if (get_area) { addr = get_area(file, addr, len, pgoff, flags); - } else if (IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE)) { + } else if (IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE) + && IS_ALIGNED(len, PMD_SIZE)) { /* Ensures that larger anonymous mappings are THP aligned. */ addr = thp_get_unmapped_area_vmflags(file, addr, len, pgoff, flags, vm_flags); -- 2.51.0 From bcfe43f0ea77c42c2154fb79b99b7d1d82ac3231 Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Mon, 28 Oct 2024 16:35:07 +0000 Subject: [PATCH 03/16] drm/imx/dcss: include drm/drm_bridge.h header Compile-testing random configurations leads to failures in dcss-kms.c from a missing declaration: drivers/gpu/drm/imx/dcss/dcss-kms.c:95:8: error: use of undeclared identifier 'drm_bridge_attach' 95 | ret = drm_bridge_attach(encoder, bridge, NULL, | ^ drivers/gpu/drm/imx/dcss/dcss-kms.c:96:5: error: use of undeclared identifier 'DRM_BRIDGE_ATTACH_NO_CONNECTOR' 96 | DRM_BRIDGE_ATTACH_NO_CONNECTOR); | ^ Include the header directly. Fixes: 004555a18d57 ("drm/imx/dcss: Allow build with COMPILE_TEST=y") Signed-off-by: Arnd Bergmann Signed-off-by: Dave Airlie Link: https://patchwork.freedesktop.org/patch/msgid/20241028163527.2425783-1-arnd@kernel.org --- drivers/gpu/drm/imx/dcss/dcss-kms.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/imx/dcss/dcss-kms.c b/drivers/gpu/drm/imx/dcss/dcss-kms.c index 3ec721afc30c..63a335c62296 100644 --- a/drivers/gpu/drm/imx/dcss/dcss-kms.c +++ b/drivers/gpu/drm/imx/dcss/dcss-kms.c @@ -5,6 +5,7 @@ #include #include +#include #include #include #include -- 2.51.0 From 071b24b54d2d05fbf39ddbb27dee08abd1d713f3 Mon Sep 17 00:00:00 2001 From: Dmitry Torokhov Date: Sun, 27 Oct 2024 22:31:15 -0700 Subject: [PATCH 04/16] Input: fix regression when re-registering input handlers Commit d469647bafd9 ("Input: simplify event handling logic") introduced code that would set handler->events() method to either input_handler_events_filter() or input_handler_events_default() or input_handler_events_null(), depending on the kind of input handler (a filter or a regular one) we are dealing with. Unfortunately this breaks cases when we try to re-register the same filter (as is the case with sysrq handler): after initial registration the handler will have 2 event handling methods defined, and will run afoul of the check in input_handler_check_methods(): input: input_handler_check_methods: only one event processing method can be defined (sysrq) sysrq: Failed to register input handler, error -22 Fix this by adding handle_events() method to input_handle structure and setting it up when registering a new input handle according to event handling methods defined in associated input_handler structure, thus avoiding modifying the input_handler structure. Reported-by: "Ned T. Crigler" Reported-by: Christian Heusel Tested-by: "Ned T. Crigler" Tested-by: Peter Seiderer Fixes: d469647bafd9 ("Input: simplify event handling logic") Link: https://lore.kernel.org/r/Zx2iQp6csn42PJA7@xavtug Cc: stable@vger.kernel.org Signed-off-by: Dmitry Torokhov --- drivers/input/input.c | 134 +++++++++++++++++++++++------------------- include/linux/input.h | 10 +++- 2 files changed, 82 insertions(+), 62 deletions(-) diff --git a/drivers/input/input.c b/drivers/input/input.c index 3c321671793f..3d2cc13e1f32 100644 --- a/drivers/input/input.c +++ b/drivers/input/input.c @@ -119,12 +119,12 @@ static void input_pass_values(struct input_dev *dev, handle = rcu_dereference(dev->grab); if (handle) { - count = handle->handler->events(handle, vals, count); + count = handle->handle_events(handle, vals, count); } else { list_for_each_entry_rcu(handle, &dev->h_list, d_node) if (handle->open) { - count = handle->handler->events(handle, vals, - count); + count = handle->handle_events(handle, vals, + count); if (!count) break; } @@ -2537,57 +2537,6 @@ static int input_handler_check_methods(const struct input_handler *handler) return 0; } -/* - * An implementation of input_handler's events() method that simply - * invokes handler->event() method for each event one by one. - */ -static unsigned int input_handler_events_default(struct input_handle *handle, - struct input_value *vals, - unsigned int count) -{ - struct input_handler *handler = handle->handler; - struct input_value *v; - - for (v = vals; v != vals + count; v++) - handler->event(handle, v->type, v->code, v->value); - - return count; -} - -/* - * An implementation of input_handler's events() method that invokes - * handler->filter() method for each event one by one and removes events - * that were filtered out from the "vals" array. - */ -static unsigned int input_handler_events_filter(struct input_handle *handle, - struct input_value *vals, - unsigned int count) -{ - struct input_handler *handler = handle->handler; - struct input_value *end = vals; - struct input_value *v; - - for (v = vals; v != vals + count; v++) { - if (handler->filter(handle, v->type, v->code, v->value)) - continue; - if (end != v) - *end = *v; - end++; - } - - return end - vals; -} - -/* - * An implementation of input_handler's events() method that does nothing. - */ -static unsigned int input_handler_events_null(struct input_handle *handle, - struct input_value *vals, - unsigned int count) -{ - return count; -} - /** * input_register_handler - register a new input handler * @handler: handler to be registered @@ -2607,13 +2556,6 @@ int input_register_handler(struct input_handler *handler) INIT_LIST_HEAD(&handler->h_list); - if (handler->filter) - handler->events = input_handler_events_filter; - else if (handler->event) - handler->events = input_handler_events_default; - else if (!handler->events) - handler->events = input_handler_events_null; - error = mutex_lock_interruptible(&input_mutex); if (error) return error; @@ -2687,6 +2629,75 @@ int input_handler_for_each_handle(struct input_handler *handler, void *data, } EXPORT_SYMBOL(input_handler_for_each_handle); +/* + * An implementation of input_handle's handle_events() method that simply + * invokes handler->event() method for each event one by one. + */ +static unsigned int input_handle_events_default(struct input_handle *handle, + struct input_value *vals, + unsigned int count) +{ + struct input_handler *handler = handle->handler; + struct input_value *v; + + for (v = vals; v != vals + count; v++) + handler->event(handle, v->type, v->code, v->value); + + return count; +} + +/* + * An implementation of input_handle's handle_events() method that invokes + * handler->filter() method for each event one by one and removes events + * that were filtered out from the "vals" array. + */ +static unsigned int input_handle_events_filter(struct input_handle *handle, + struct input_value *vals, + unsigned int count) +{ + struct input_handler *handler = handle->handler; + struct input_value *end = vals; + struct input_value *v; + + for (v = vals; v != vals + count; v++) { + if (handler->filter(handle, v->type, v->code, v->value)) + continue; + if (end != v) + *end = *v; + end++; + } + + return end - vals; +} + +/* + * An implementation of input_handle's handle_events() method that does nothing. + */ +static unsigned int input_handle_events_null(struct input_handle *handle, + struct input_value *vals, + unsigned int count) +{ + return count; +} + +/* + * Sets up appropriate handle->event_handler based on the input_handler + * associated with the handle. + */ +static void input_handle_setup_event_handler(struct input_handle *handle) +{ + struct input_handler *handler = handle->handler; + + if (handler->filter) + handle->handle_events = input_handle_events_filter; + else if (handler->event) + handle->handle_events = input_handle_events_default; + else if (handler->events) + handle->handle_events = handler->events; + else + handle->handle_events = input_handle_events_null; +} + /** * input_register_handle - register a new input handle * @handle: handle to register @@ -2704,6 +2715,7 @@ int input_register_handle(struct input_handle *handle) struct input_dev *dev = handle->dev; int error; + input_handle_setup_event_handler(handle); /* * We take dev->mutex here to prevent race with * input_release_device(). diff --git a/include/linux/input.h b/include/linux/input.h index 89a0be6ee0e2..cd866b020a01 100644 --- a/include/linux/input.h +++ b/include/linux/input.h @@ -339,12 +339,16 @@ struct input_handler { * @name: name given to the handle by handler that created it * @dev: input device the handle is attached to * @handler: handler that works with the device through this handle + * @handle_events: event sequence handler. It is set up by the input core + * according to event handling method specified in the @handler. See + * input_handle_setup_event_handler(). + * This method is being called by the input core with interrupts disabled + * and dev->event_lock spinlock held and so it may not sleep. * @d_node: used to put the handle on device's list of attached handles * @h_node: used to put the handle on handler's list of handles from which * it gets events */ struct input_handle { - void *private; int open; @@ -353,6 +357,10 @@ struct input_handle { struct input_dev *dev; struct input_handler *handler; + unsigned int (*handle_events)(struct input_handle *handle, + struct input_value *vals, + unsigned int count); + struct list_head d_node; struct list_head h_node; }; -- 2.51.0 From 2e766a1f5f94a142d9a906c9411d0f6101c4c721 Mon Sep 17 00:00:00 2001 From: Masahiro Yamada Date: Sun, 3 Nov 2024 21:46:50 +0900 Subject: [PATCH 05/16] modpost: fix acpi MODULE_DEVICE_TABLE built with mismatched endianness When CONFIG_SATA_AHCI_PLATFORM=m, modpost outputs incorect acpi MODULE_ALIAS() if the endianness of the target and the build machine do not match. When the endianness of the target kernel and the build machine match, the output is correct: $ grep 'MODULE_ALIAS("acpi' drivers/ata/ahci_platform.mod.c MODULE_ALIAS("acpi*:APMC0D33:*"); MODULE_ALIAS("acpi*:010601:*"); However, when building a little-endian kernel on a big-endian machine (or vice versa), the output is incorrect: $ grep 'MODULE_ALIAS("acpi' drivers/ata/ahci_platform.mod.c MODULE_ALIAS("acpi*:APMC0D33:*"); MODULE_ALIAS("acpi*:0601??:*"); The 'cls' and 'cls_msk' fields are 32-bit. DEF_FIELD() must be used instead of DEF_FIELD_ADDR() to correctly handle endianness of these 32-bit fields. The check 'if (cls)' was unnecessary; it never became NULL, as it was the pointer to 'symval' plus the offset to the 'cls' field. Fixes: 26095a01d359 ("ACPI / scan: Add support for ACPI _CLS device matching") Signed-off-by: Masahiro Yamada --- scripts/mod/file2alias.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/scripts/mod/file2alias.c b/scripts/mod/file2alias.c index 99dce93a4188..16154449dde1 100644 --- a/scripts/mod/file2alias.c +++ b/scripts/mod/file2alias.c @@ -567,12 +567,12 @@ static int do_acpi_entry(const char *filename, void *symval, char *alias) { DEF_FIELD_ADDR(symval, acpi_device_id, id); - DEF_FIELD_ADDR(symval, acpi_device_id, cls); - DEF_FIELD_ADDR(symval, acpi_device_id, cls_msk); + DEF_FIELD(symval, acpi_device_id, cls); + DEF_FIELD(symval, acpi_device_id, cls_msk); if (id && strlen((const char *)*id)) sprintf(alias, "acpi*:%s:*", *id); - else if (cls) { + else { int i, byte_shift, cnt = 0; unsigned int msk; @@ -580,10 +580,10 @@ static int do_acpi_entry(const char *filename, cnt = 6; for (i = 1; i <= 3; i++) { byte_shift = 8 * (3-i); - msk = (*cls_msk >> byte_shift) & 0xFF; + msk = (cls_msk >> byte_shift) & 0xFF; if (msk) sprintf(&alias[cnt], "%02x", - (*cls >> byte_shift) & 0xFF); + (cls >> byte_shift) & 0xFF); else sprintf(&alias[cnt], "??"); cnt += 2; -- 2.51.0 From 77dc55a978e69625f9718460012e5ef0172dc4de Mon Sep 17 00:00:00 2001 From: Masahiro Yamada Date: Sun, 3 Nov 2024 21:52:57 +0900 Subject: [PATCH 06/16] modpost: fix input MODULE_DEVICE_TABLE() built for 64-bit on 32-bit host When building a 64-bit kernel on a 32-bit build host, incorrect input MODULE_ALIAS() entries may be generated. For example, when compiling a 64-bit kernel with CONFIG_INPUT_MOUSEDEV=m on a 64-bit build machine, you will get the correct output: $ grep MODULE_ALIAS drivers/input/mousedev.mod.c MODULE_ALIAS("input:b*v*p*e*-e*1,*2,*k*110,*r*0,*1,*a*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*1,*2,*k*r*8,*a*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*1,*3,*k*14A,*r*a*0,*1,*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*1,*3,*k*145,*r*a*0,*1,*18,*1C,*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*1,*3,*k*110,*r*a*0,*1,*m*l*s*f*w*"); However, building the same kernel on a 32-bit machine results in incorrect output: $ grep MODULE_ALIAS drivers/input/mousedev.mod.c MODULE_ALIAS("input:b*v*p*e*-e*1,*2,*k*110,*130,*r*0,*1,*a*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*1,*2,*k*r*8,*a*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*1,*3,*k*14A,*16A,*r*a*0,*1,*20,*21,*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*1,*3,*k*145,*165,*r*a*0,*1,*18,*1C,*20,*21,*38,*3C,*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*1,*3,*k*110,*130,*r*a*0,*1,*20,*21,*m*l*s*f*w*"); A similar issue occurs with CONFIG_INPUT_JOYDEV=m. On a 64-bit build machine, the output is: $ grep MODULE_ALIAS drivers/input/joydev.mod.c MODULE_ALIAS("input:b*v*p*e*-e*3,*k*r*a*0,*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*3,*k*r*a*2,*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*3,*k*r*a*8,*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*3,*k*r*a*6,*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*1,*k*120,*r*a*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*1,*k*130,*r*a*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*1,*k*2C0,*r*a*m*l*s*f*w*"); However, on a 32-bit machine, the output is incorrect: $ grep MODULE_ALIAS drivers/input/joydev.mod.c MODULE_ALIAS("input:b*v*p*e*-e*3,*k*r*a*0,*20,*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*3,*k*r*a*2,*22,*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*3,*k*r*a*8,*28,*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*3,*k*r*a*6,*26,*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*1,*k*11F,*13F,*r*a*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*1,*k*11F,*13F,*r*a*m*l*s*f*w*"); MODULE_ALIAS("input:b*v*p*e*-e*1,*k*2C0,*2E0,*r*a*m*l*s*f*w*"); When building a 64-bit kernel, BITS_PER_LONG is defined as 64. However, on a 32-bit build machine, the constant 1L is a signed 32-bit value. Left-shifting it beyond 32 bits causes wraparound, and shifting by 31 or 63 bits makes it a negative value. The fix in commit e0e92632715f ("[PATCH] PATCH: 1 line 2.6.18 bugfix: modpost-64bit-fix.patch") is incorrect; it only addresses cases where a 64-bit kernel is built on a 64-bit build machine, overlooking cases on a 32-bit build machine. Using 1ULL ensures a 64-bit width on both 32-bit and 64-bit machines, avoiding the wraparound issue. Fixes: e0e92632715f ("[PATCH] PATCH: 1 line 2.6.18 bugfix: modpost-64bit-fix.patch") Signed-off-by: Masahiro Yamada --- scripts/mod/file2alias.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/scripts/mod/file2alias.c b/scripts/mod/file2alias.c index 16154449dde1..c4cc11aa558f 100644 --- a/scripts/mod/file2alias.c +++ b/scripts/mod/file2alias.c @@ -743,7 +743,7 @@ static void do_input(char *alias, for (i = min / BITS_PER_LONG; i < max / BITS_PER_LONG + 1; i++) arr[i] = TO_NATIVE(arr[i]); for (i = min; i < max; i++) - if (arr[i / BITS_PER_LONG] & (1L << (i%BITS_PER_LONG))) + if (arr[i / BITS_PER_LONG] & (1ULL << (i%BITS_PER_LONG))) sprintf(alias + strlen(alias), "%X,*", i); } -- 2.51.0 From ddd6d8e975b171ea3f63a011a75820883ff0d479 Mon Sep 17 00:00:00 2001 From: Yu Zhao Date: Sat, 19 Oct 2024 01:29:38 +0000 Subject: [PATCH 07/16] mm: multi-gen LRU: remove MM_LEAF_OLD and MM_NONLEAF_TOTAL stats Patch series "mm: multi-gen LRU: Have secondary MMUs participate in MM_WALK". Today, the MM_WALK capability causes MGLRU to clear the young bit from PMDs and PTEs during the page table walk before eviction, but MGLRU does not call the clear_young() MMU notifier in this case. By not calling this notifier, the MM walk takes less time/CPU, but it causes pages that are accessed mostly through KVM / secondary MMUs to appear younger than they should be. We do call the clear_young() notifier today, but only when attempting to evict the page, so we end up clearing young/accessed information less frequently for secondary MMUs than for mm PTEs, and therefore they appear younger and are less likely to be evicted. Therefore, memory that is *not* being accessed mostly by KVM will be evicted *more* frequently, worsening performance. ChromeOS observed a tab-open latency regression when enabling MGLRU with a setup that involved running a VM: Tab-open latency histogram (ms) Version p50 mean p95 p99 max base 1315 1198 2347 3454 10319 mglru 2559 1311 7399 12060 43758 fix 1119 926 2470 4211 6947 This series replaces the final non-selftest patchs from this series[1], which introduced a similar change (and a new MMU notifier) with KVM optimizations. I'll send a separate series (to Sean and Paolo) for the KVM optimizations. This series also makes proactive reclaim with MGLRU possible for KVM memory. I have verified that this functions correctly with the selftest from [1], but given that that test is a KVM selftest, I'll send it with the rest of the KVM optimizations later. Andrew, let me know if you'd like to take the test now anyway. [1]: https://lore.kernel.org/linux-mm/20240926013506.860253-18-jthoughton@google.com/ This patch (of 2): The removed stats, MM_LEAF_OLD and MM_NONLEAF_TOTAL, are not very helpful and become more complicated to properly compute when adding test/clear_young() notifiers in MGLRU's mm walk. Link: https://lkml.kernel.org/r/20241019012940.3656292-1-jthoughton@google.com Link: https://lkml.kernel.org/r/20241019012940.3656292-2-jthoughton@google.com Fixes: bd74fdaea146 ("mm: multi-gen LRU: support page table walks") Signed-off-by: Yu Zhao Signed-off-by: James Houghton Cc: Axel Rasmussen Cc: David Matlack Cc: David Rientjes Cc: David Stevens Cc: Oliver Upton Cc: Paolo Bonzini Cc: Sean Christopherson Cc: Wei Xu Cc: Signed-off-by: Andrew Morton --- include/linux/mmzone.h | 2 -- mm/vmscan.c | 14 +++++--------- 2 files changed, 5 insertions(+), 11 deletions(-) diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h index 17506e4a2835..9342e5692dab 100644 --- a/include/linux/mmzone.h +++ b/include/linux/mmzone.h @@ -458,9 +458,7 @@ struct lru_gen_folio { enum { MM_LEAF_TOTAL, /* total leaf entries */ - MM_LEAF_OLD, /* old leaf entries */ MM_LEAF_YOUNG, /* young leaf entries */ - MM_NONLEAF_TOTAL, /* total non-leaf entries */ MM_NONLEAF_FOUND, /* non-leaf entries found in Bloom filters */ MM_NONLEAF_ADDED, /* non-leaf entries added to Bloom filters */ NR_MM_STATS diff --git a/mm/vmscan.c b/mm/vmscan.c index eb4e8440c507..4f1d33e4b360 100644 --- a/mm/vmscan.c +++ b/mm/vmscan.c @@ -3399,7 +3399,6 @@ restart: continue; if (!pte_young(ptent)) { - walk->mm_stats[MM_LEAF_OLD]++; continue; } @@ -3552,7 +3551,6 @@ restart: walk->mm_stats[MM_LEAF_TOTAL]++; if (!pmd_young(val)) { - walk->mm_stats[MM_LEAF_OLD]++; continue; } @@ -3564,8 +3562,6 @@ restart: continue; } - walk->mm_stats[MM_NONLEAF_TOTAL]++; - if (!walk->force_scan && should_clear_pmd_young()) { if (!pmd_young(val)) continue; @@ -5254,11 +5250,11 @@ static void lru_gen_seq_show_full(struct seq_file *m, struct lruvec *lruvec, for (tier = 0; tier < MAX_NR_TIERS; tier++) { seq_printf(m, " %10d", tier); for (type = 0; type < ANON_AND_FILE; type++) { - const char *s = " "; + const char *s = "xxx"; unsigned long n[3] = {}; if (seq == max_seq) { - s = "RT "; + s = "RTx"; n[0] = READ_ONCE(lrugen->avg_refaulted[type][tier]); n[1] = READ_ONCE(lrugen->avg_total[type][tier]); } else if (seq == min_seq[type] || NR_HIST_GENS > 1) { @@ -5280,14 +5276,14 @@ static void lru_gen_seq_show_full(struct seq_file *m, struct lruvec *lruvec, seq_puts(m, " "); for (i = 0; i < NR_MM_STATS; i++) { - const char *s = " "; + const char *s = "xxxx"; unsigned long n = 0; if (seq == max_seq && NR_HIST_GENS == 1) { - s = "LOYNFA"; + s = "TYFA"; n = READ_ONCE(mm_state->stats[hist][i]); } else if (seq != max_seq && NR_HIST_GENS > 1) { - s = "loynfa"; + s = "tyfa"; n = READ_ONCE(mm_state->stats[hist][i]); } -- 2.51.0 From 1d4832becdc2cdb2cffe2a6050c9d9fd8ff1c58c Mon Sep 17 00:00:00 2001 From: Yu Zhao Date: Sat, 19 Oct 2024 01:29:39 +0000 Subject: [PATCH 08/16] mm: multi-gen LRU: use {ptep,pmdp}_clear_young_notify() When the MM_WALK capability is enabled, memory that is mostly accessed by a VM appears younger than it really is, therefore this memory will be less likely to be evicted. Therefore, the presence of a running VM can significantly increase swap-outs for non-VM memory, regressing the performance for the rest of the system. Fix this regression by always calling {ptep,pmdp}_clear_young_notify() whenever we clear the young bits on PMDs/PTEs. [jthoughton@google.com: fix link-time error] Link: https://lkml.kernel.org/r/20241019012940.3656292-3-jthoughton@google.com Fixes: bd74fdaea146 ("mm: multi-gen LRU: support page table walks") Signed-off-by: Yu Zhao Signed-off-by: James Houghton Reported-by: David Stevens Cc: Axel Rasmussen Cc: David Matlack Cc: David Rientjes Cc: Oliver Upton Cc: Paolo Bonzini Cc: Sean Christopherson Cc: Wei Xu Cc: Cc: kernel test robot Signed-off-by: Andrew Morton --- include/linux/mmzone.h | 5 ++- mm/rmap.c | 9 ++--- mm/vmscan.c | 88 +++++++++++++++++++++++------------------- 3 files changed, 55 insertions(+), 47 deletions(-) diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h index 9342e5692dab..5b1c984daf45 100644 --- a/include/linux/mmzone.h +++ b/include/linux/mmzone.h @@ -555,7 +555,7 @@ struct lru_gen_memcg { void lru_gen_init_pgdat(struct pglist_data *pgdat); void lru_gen_init_lruvec(struct lruvec *lruvec); -void lru_gen_look_around(struct page_vma_mapped_walk *pvmw); +bool lru_gen_look_around(struct page_vma_mapped_walk *pvmw); void lru_gen_init_memcg(struct mem_cgroup *memcg); void lru_gen_exit_memcg(struct mem_cgroup *memcg); @@ -574,8 +574,9 @@ static inline void lru_gen_init_lruvec(struct lruvec *lruvec) { } -static inline void lru_gen_look_around(struct page_vma_mapped_walk *pvmw) +static inline bool lru_gen_look_around(struct page_vma_mapped_walk *pvmw) { + return false; } static inline void lru_gen_init_memcg(struct mem_cgroup *memcg) diff --git a/mm/rmap.c b/mm/rmap.c index a8797d1b3d49..73d5998677d4 100644 --- a/mm/rmap.c +++ b/mm/rmap.c @@ -885,13 +885,10 @@ static bool folio_referenced_one(struct folio *folio, return false; } - if (pvmw.pte) { - if (lru_gen_enabled() && - pte_young(ptep_get(pvmw.pte))) { - lru_gen_look_around(&pvmw); + if (lru_gen_enabled() && pvmw.pte) { + if (lru_gen_look_around(&pvmw)) referenced++; - } - + } else if (pvmw.pte) { if (ptep_clear_flush_young_notify(vma, address, pvmw.pte)) referenced++; diff --git a/mm/vmscan.c b/mm/vmscan.c index 4f1d33e4b360..ddaaff67642e 100644 --- a/mm/vmscan.c +++ b/mm/vmscan.c @@ -56,6 +56,7 @@ #include #include #include +#include #include #include @@ -3294,7 +3295,8 @@ static bool get_next_vma(unsigned long mask, unsigned long size, struct mm_walk return false; } -static unsigned long get_pte_pfn(pte_t pte, struct vm_area_struct *vma, unsigned long addr) +static unsigned long get_pte_pfn(pte_t pte, struct vm_area_struct *vma, unsigned long addr, + struct pglist_data *pgdat) { unsigned long pfn = pte_pfn(pte); @@ -3306,13 +3308,20 @@ static unsigned long get_pte_pfn(pte_t pte, struct vm_area_struct *vma, unsigned if (WARN_ON_ONCE(pte_devmap(pte) || pte_special(pte))) return -1; + if (!pte_young(pte) && !mm_has_notifiers(vma->vm_mm)) + return -1; + if (WARN_ON_ONCE(!pfn_valid(pfn))) return -1; + if (pfn < pgdat->node_start_pfn || pfn >= pgdat_end_pfn(pgdat)) + return -1; + return pfn; } -static unsigned long get_pmd_pfn(pmd_t pmd, struct vm_area_struct *vma, unsigned long addr) +static unsigned long get_pmd_pfn(pmd_t pmd, struct vm_area_struct *vma, unsigned long addr, + struct pglist_data *pgdat) { unsigned long pfn = pmd_pfn(pmd); @@ -3324,9 +3333,15 @@ static unsigned long get_pmd_pfn(pmd_t pmd, struct vm_area_struct *vma, unsigned if (WARN_ON_ONCE(pmd_devmap(pmd))) return -1; + if (!pmd_young(pmd) && !mm_has_notifiers(vma->vm_mm)) + return -1; + if (WARN_ON_ONCE(!pfn_valid(pfn))) return -1; + if (pfn < pgdat->node_start_pfn || pfn >= pgdat_end_pfn(pgdat)) + return -1; + return pfn; } @@ -3335,10 +3350,6 @@ static struct folio *get_pfn_folio(unsigned long pfn, struct mem_cgroup *memcg, { struct folio *folio; - /* try to avoid unnecessary memory loads */ - if (pfn < pgdat->node_start_pfn || pfn >= pgdat_end_pfn(pgdat)) - return NULL; - folio = pfn_folio(pfn); if (folio_nid(folio) != pgdat->node_id) return NULL; @@ -3394,20 +3405,16 @@ restart: total++; walk->mm_stats[MM_LEAF_TOTAL]++; - pfn = get_pte_pfn(ptent, args->vma, addr); + pfn = get_pte_pfn(ptent, args->vma, addr, pgdat); if (pfn == -1) continue; - if (!pte_young(ptent)) { - continue; - } - folio = get_pfn_folio(pfn, memcg, pgdat, walk->can_swap); if (!folio) continue; - if (!ptep_test_and_clear_young(args->vma, addr, pte + i)) - VM_WARN_ON_ONCE(true); + if (!ptep_clear_young_notify(args->vma, addr, pte + i)) + continue; young++; walk->mm_stats[MM_LEAF_YOUNG]++; @@ -3473,21 +3480,25 @@ static void walk_pmd_range_locked(pud_t *pud, unsigned long addr, struct vm_area /* don't round down the first address */ addr = i ? (*first & PMD_MASK) + i * PMD_SIZE : *first; - pfn = get_pmd_pfn(pmd[i], vma, addr); - if (pfn == -1) + if (!pmd_present(pmd[i])) goto next; if (!pmd_trans_huge(pmd[i])) { - if (!walk->force_scan && should_clear_pmd_young()) + if (!walk->force_scan && should_clear_pmd_young() && + !mm_has_notifiers(args->mm)) pmdp_test_and_clear_young(vma, addr, pmd + i); goto next; } + pfn = get_pmd_pfn(pmd[i], vma, addr, pgdat); + if (pfn == -1) + goto next; + folio = get_pfn_folio(pfn, memcg, pgdat, walk->can_swap); if (!folio) goto next; - if (!pmdp_test_and_clear_young(vma, addr, pmd + i)) + if (!pmdp_clear_young_notify(vma, addr, pmd + i)) goto next; walk->mm_stats[MM_LEAF_YOUNG]++; @@ -3545,24 +3556,18 @@ restart: } if (pmd_trans_huge(val)) { - unsigned long pfn = pmd_pfn(val); struct pglist_data *pgdat = lruvec_pgdat(walk->lruvec); + unsigned long pfn = get_pmd_pfn(val, vma, addr, pgdat); walk->mm_stats[MM_LEAF_TOTAL]++; - if (!pmd_young(val)) { - continue; - } - - /* try to avoid unnecessary memory loads */ - if (pfn < pgdat->node_start_pfn || pfn >= pgdat_end_pfn(pgdat)) - continue; - - walk_pmd_range_locked(pud, addr, vma, args, bitmap, &first); + if (pfn != -1) + walk_pmd_range_locked(pud, addr, vma, args, bitmap, &first); continue; } - if (!walk->force_scan && should_clear_pmd_young()) { + if (!walk->force_scan && should_clear_pmd_young() && + !mm_has_notifiers(args->mm)) { if (!pmd_young(val)) continue; @@ -4036,13 +4041,13 @@ static void lru_gen_age_node(struct pglist_data *pgdat, struct scan_control *sc) * the PTE table to the Bloom filter. This forms a feedback loop between the * eviction and the aging. */ -void lru_gen_look_around(struct page_vma_mapped_walk *pvmw) +bool lru_gen_look_around(struct page_vma_mapped_walk *pvmw) { int i; unsigned long start; unsigned long end; struct lru_gen_mm_walk *walk; - int young = 0; + int young = 1; pte_t *pte = pvmw->pte; unsigned long addr = pvmw->address; struct vm_area_struct *vma = pvmw->vma; @@ -4058,12 +4063,15 @@ void lru_gen_look_around(struct page_vma_mapped_walk *pvmw) lockdep_assert_held(pvmw->ptl); VM_WARN_ON_ONCE_FOLIO(folio_test_lru(folio), folio); + if (!ptep_clear_young_notify(vma, addr, pte)) + return false; + if (spin_is_contended(pvmw->ptl)) - return; + return true; /* exclude special VMAs containing anon pages from COW */ if (vma->vm_flags & VM_SPECIAL) - return; + return true; /* avoid taking the LRU lock under the PTL when possible */ walk = current->reclaim_state ? current->reclaim_state->mm_walk : NULL; @@ -4071,6 +4079,9 @@ void lru_gen_look_around(struct page_vma_mapped_walk *pvmw) start = max(addr & PMD_MASK, vma->vm_start); end = min(addr | ~PMD_MASK, vma->vm_end - 1) + 1; + if (end - start == PAGE_SIZE) + return true; + if (end - start > MIN_LRU_BATCH * PAGE_SIZE) { if (addr - start < MIN_LRU_BATCH * PAGE_SIZE / 2) end = start + MIN_LRU_BATCH * PAGE_SIZE; @@ -4084,7 +4095,7 @@ void lru_gen_look_around(struct page_vma_mapped_walk *pvmw) /* folio_update_gen() requires stable folio_memcg() */ if (!mem_cgroup_trylock_pages(memcg)) - return; + return true; arch_enter_lazy_mmu_mode(); @@ -4094,19 +4105,16 @@ void lru_gen_look_around(struct page_vma_mapped_walk *pvmw) unsigned long pfn; pte_t ptent = ptep_get(pte + i); - pfn = get_pte_pfn(ptent, vma, addr); + pfn = get_pte_pfn(ptent, vma, addr, pgdat); if (pfn == -1) continue; - if (!pte_young(ptent)) - continue; - folio = get_pfn_folio(pfn, memcg, pgdat, can_swap); if (!folio) continue; - if (!ptep_test_and_clear_young(vma, addr, pte + i)) - VM_WARN_ON_ONCE(true); + if (!ptep_clear_young_notify(vma, addr, pte + i)) + continue; young++; @@ -4136,6 +4144,8 @@ void lru_gen_look_around(struct page_vma_mapped_walk *pvmw) /* feedback from rmap walkers to page table walkers */ if (mm_state && suitable_to_scan(i, young)) update_bloom_filter(mm_state, max_seq, pvmw->pmd); + + return true; } /****************************************************************************** -- 2.51.0 From 59b723cd2adbac2a34fc8e12c74ae26ae45bf230 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Sun, 3 Nov 2024 14:05:52 -1000 Subject: [PATCH 09/16] Linux 6.12-rc6 --- Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Makefile b/Makefile index 5e04e4abffd8..b8efbfe9da94 100644 --- a/Makefile +++ b/Makefile @@ -2,7 +2,7 @@ VERSION = 6 PATCHLEVEL = 12 SUBLEVEL = 0 -EXTRAVERSION = -rc5 +EXTRAVERSION = -rc6 NAME = Baby Opossum Posse # *DOCUMENTATION* -- 2.51.0 From bf56dc2138b92ffbf68e2e36b82d75b0b105564d Mon Sep 17 00:00:00 2001 From: Shen Lichuan Date: Mon, 9 Sep 2024 16:06:20 +0800 Subject: [PATCH 10/16] drm/exynos: gsc: Fix typo in comment Replace 'initailization' with 'initialization' in the comment. Signed-off-by: Shen Lichuan Signed-off-by: Inki Dae --- drivers/gpu/drm/exynos/exynos_drm_gsc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/exynos/exynos_drm_gsc.c b/drivers/gpu/drm/exynos/exynos_drm_gsc.c index 59fa22050717..1ae90ef1fc23 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_gsc.c +++ b/drivers/gpu/drm/exynos/exynos_drm_gsc.c @@ -1286,7 +1286,7 @@ static int gsc_probe(struct platform_device *pdev) return ret; } - /* context initailization */ + /* context initialization */ ctx->id = pdev->id; platform_set_drvdata(pdev, ctx); -- 2.51.0 From c941a4902f6fa95a5d0bc0d74ee718c347defca3 Mon Sep 17 00:00:00 2001 From: Kwanghoon Son Date: Thu, 26 Sep 2024 14:25:39 +0900 Subject: [PATCH 11/16] drm/exynos: remove unused prototype for crtc exynos_drm_crtc_wait_pending_update, exynos_drm_crtc_finish_update are not used anymore. Signed-off-by: Kwanghoon Son Signed-off-by: Inki Dae --- drivers/gpu/drm/exynos/exynos_drm_crtc.h | 3 --- 1 file changed, 3 deletions(-) diff --git a/drivers/gpu/drm/exynos/exynos_drm_crtc.h b/drivers/gpu/drm/exynos/exynos_drm_crtc.h index 0ed4f2b8595a..1815374c38df 100644 --- a/drivers/gpu/drm/exynos/exynos_drm_crtc.h +++ b/drivers/gpu/drm/exynos/exynos_drm_crtc.h @@ -19,9 +19,6 @@ struct exynos_drm_crtc *exynos_drm_crtc_create(struct drm_device *drm_dev, enum exynos_drm_output_type out_type, const struct exynos_drm_crtc_ops *ops, void *context); -void exynos_drm_crtc_wait_pending_update(struct exynos_drm_crtc *exynos_crtc); -void exynos_drm_crtc_finish_update(struct exynos_drm_crtc *exynos_crtc, - struct exynos_drm_plane *exynos_plane); /* This function gets crtc device matched with out_type. */ struct exynos_drm_crtc *exynos_drm_crtc_get_by_type(struct drm_device *drm_dev, -- 2.51.0 From d31bbacf783daf1e71fbe5c68df93550c446bf44 Mon Sep 17 00:00:00 2001 From: Kaustabh Chakraborty Date: Fri, 1 Nov 2024 11:27:27 +0900 Subject: [PATCH 12/16] drm/exynos: exynos7_drm_decon: fix uninitialized crtc reference in functions Modify the functions to accept a pointer to struct decon_context instead. Signed-off-by: Kaustabh Chakraborty Signed-off-by: Inki Dae --- drivers/gpu/drm/exynos/exynos7_drm_decon.c | 11 ++++------- 1 file changed, 4 insertions(+), 7 deletions(-) diff --git a/drivers/gpu/drm/exynos/exynos7_drm_decon.c b/drivers/gpu/drm/exynos/exynos7_drm_decon.c index 0d185c0564b9..e994779694f0 100644 --- a/drivers/gpu/drm/exynos/exynos7_drm_decon.c +++ b/drivers/gpu/drm/exynos/exynos7_drm_decon.c @@ -81,10 +81,8 @@ static const enum drm_plane_type decon_win_types[WINDOWS_NR] = { DRM_PLANE_TYPE_CURSOR, }; -static void decon_wait_for_vblank(struct exynos_drm_crtc *crtc) +static void decon_wait_for_vblank(struct decon_context *ctx) { - struct decon_context *ctx = crtc->ctx; - if (ctx->suspended) return; @@ -100,9 +98,8 @@ static void decon_wait_for_vblank(struct exynos_drm_crtc *crtc) DRM_DEV_DEBUG_KMS(ctx->dev, "vblank wait timed out.\n"); } -static void decon_clear_channels(struct exynos_drm_crtc *crtc) +static void decon_clear_channels(struct decon_context *ctx) { - struct decon_context *ctx = crtc->ctx; unsigned int win, ch_enabled = 0; /* Check if any channel is enabled. */ @@ -118,7 +115,7 @@ static void decon_clear_channels(struct exynos_drm_crtc *crtc) /* Wait for vsync, as disable channel takes effect at next vsync */ if (ch_enabled) - decon_wait_for_vblank(ctx->crtc); + decon_wait_for_vblank(ctx); } static int decon_ctx_initialize(struct decon_context *ctx, @@ -126,7 +123,7 @@ static int decon_ctx_initialize(struct decon_context *ctx, { ctx->drm_dev = drm_dev; - decon_clear_channels(ctx->crtc); + decon_clear_channels(ctx); return exynos_drm_register_dma(drm_dev, ctx->dev, &ctx->dma_priv); } -- 2.51.0 From f3cb045e2603e80a1633883423b2621aad77989d Mon Sep 17 00:00:00 2001 From: Kaustabh Chakraborty Date: Fri, 1 Nov 2024 11:31:10 +0900 Subject: [PATCH 13/16] drm/exynos: exynos7_drm_decon: fix ideal_clk by converting it to Hz The clkdiv values are incorrect as ideal_clk is in kHz and the clock rate of vclk is in Hz. Multiply 1000 to ideal_clk to bring it to Hz. Signed-off-by: Kaustabh Chakraborty Signed-off-by: Inki Dae --- drivers/gpu/drm/exynos/exynos7_drm_decon.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/exynos/exynos7_drm_decon.c b/drivers/gpu/drm/exynos/exynos7_drm_decon.c index e994779694f0..76a3f4b0341d 100644 --- a/drivers/gpu/drm/exynos/exynos7_drm_decon.c +++ b/drivers/gpu/drm/exynos/exynos7_drm_decon.c @@ -137,7 +137,7 @@ static void decon_ctx_remove(struct decon_context *ctx) static u32 decon_calc_clkdiv(struct decon_context *ctx, const struct drm_display_mode *mode) { - unsigned long ideal_clk = mode->clock; + unsigned long ideal_clk = mode->clock * 1000; u32 clkdiv; /* Find the clock divider value that gets us closest to ideal_clk */ -- 2.51.0 From 5f1a453974204175f20b3788824a0fe23cc36f79 Mon Sep 17 00:00:00 2001 From: Kaustabh Chakraborty Date: Fri, 1 Nov 2024 11:39:11 +0900 Subject: [PATCH 14/16] drm/exynos: exynos7_drm_decon: properly clear channels during bind The DECON channels are not cleared properly as the windows aren't shadow protected. When accompanied with an IOMMU, it pagefaults, and the kernel panics. Implement shadow protect/unprotect, along with a standalone update, for channel clearing to properly take effect. Signed-off-by: Kaustabh Chakraborty Signed-off-by: Inki Dae --- drivers/gpu/drm/exynos/exynos7_drm_decon.c | 55 +++++++++++++--------- 1 file changed, 32 insertions(+), 23 deletions(-) diff --git a/drivers/gpu/drm/exynos/exynos7_drm_decon.c b/drivers/gpu/drm/exynos/exynos7_drm_decon.c index 76a3f4b0341d..29c048a61b0b 100644 --- a/drivers/gpu/drm/exynos/exynos7_drm_decon.c +++ b/drivers/gpu/drm/exynos/exynos7_drm_decon.c @@ -81,6 +81,28 @@ static const enum drm_plane_type decon_win_types[WINDOWS_NR] = { DRM_PLANE_TYPE_CURSOR, }; +/** + * decon_shadow_protect_win() - disable updating values from shadow registers at vsync + * + * @ctx: display and enhancement controller context + * @win: window to protect registers for + * @protect: 1 to protect (disable updates) + */ +static void decon_shadow_protect_win(struct decon_context *ctx, + unsigned int win, bool protect) +{ + u32 bits, val; + + bits = SHADOWCON_WINx_PROTECT(win); + + val = readl(ctx->regs + SHADOWCON); + if (protect) + val |= bits; + else + val &= ~bits; + writel(val, ctx->regs + SHADOWCON); +} + static void decon_wait_for_vblank(struct decon_context *ctx) { if (ctx->suspended) @@ -101,18 +123,27 @@ static void decon_wait_for_vblank(struct decon_context *ctx) static void decon_clear_channels(struct decon_context *ctx) { unsigned int win, ch_enabled = 0; + u32 val; /* Check if any channel is enabled. */ for (win = 0; win < WINDOWS_NR; win++) { - u32 val = readl(ctx->regs + WINCON(win)); + val = readl(ctx->regs + WINCON(win)); if (val & WINCONx_ENWIN) { + decon_shadow_protect_win(ctx, win, true); + val &= ~WINCONx_ENWIN; writel(val, ctx->regs + WINCON(win)); ch_enabled = 1; + + decon_shadow_protect_win(ctx, win, false); } } + val = readl(ctx->regs + DECON_UPDATE); + val |= DECON_UPDATE_STANDALONE_F; + writel(val, ctx->regs + DECON_UPDATE); + /* Wait for vsync, as disable channel takes effect at next vsync */ if (ch_enabled) decon_wait_for_vblank(ctx); @@ -340,28 +371,6 @@ static void decon_win_set_colkey(struct decon_context *ctx, unsigned int win) writel(keycon1, ctx->regs + WKEYCON1_BASE(win)); } -/** - * decon_shadow_protect_win() - disable updating values from shadow registers at vsync - * - * @ctx: display and enhancement controller context - * @win: window to protect registers for - * @protect: 1 to protect (disable updates) - */ -static void decon_shadow_protect_win(struct decon_context *ctx, - unsigned int win, bool protect) -{ - u32 bits, val; - - bits = SHADOWCON_WINx_PROTECT(win); - - val = readl(ctx->regs + SHADOWCON); - if (protect) - val |= bits; - else - val &= ~bits; - writel(val, ctx->regs + SHADOWCON); -} - static void decon_atomic_begin(struct exynos_drm_crtc *crtc) { struct decon_context *ctx = crtc->ctx; -- 2.51.0 From 53f4b30b05b2a9db6988cb71a785837ee64d2524 Mon Sep 17 00:00:00 2001 From: Kaustabh Chakraborty Date: Fri, 1 Nov 2024 11:44:42 +0900 Subject: [PATCH 15/16] drm/exynos: exynos7_drm_decon: add driver data and support for Exynos7870 Add support for Exynos 7870 DECON in the Exynos 7 DECON driver. Some Exynos 7 series SoCs (Exynos 7580 onwards) have different register values. In order to address such changes, include a driver data struct (named decon_data) so that correct base addresses and shift values can be provided. Signed-off-by: Kaustabh Chakraborty Signed-off-by: Inki Dae --- drivers/gpu/drm/exynos/exynos7_drm_decon.c | 58 ++++++++++++++++------ drivers/gpu/drm/exynos/regs-decon7.h | 15 +++--- 2 files changed, 51 insertions(+), 22 deletions(-) diff --git a/drivers/gpu/drm/exynos/exynos7_drm_decon.c b/drivers/gpu/drm/exynos/exynos7_drm_decon.c index 29c048a61b0b..c65364087fac 100644 --- a/drivers/gpu/drm/exynos/exynos7_drm_decon.c +++ b/drivers/gpu/drm/exynos/exynos7_drm_decon.c @@ -37,6 +37,24 @@ #define WINDOWS_NR 2 +struct decon_data { + unsigned int vidw_buf_start_base; + unsigned int shadowcon_win_protect_shift; + unsigned int wincon_burstlen_shift; +}; + +static struct decon_data exynos7_decon_data = { + .vidw_buf_start_base = 0x80, + .shadowcon_win_protect_shift = 10, + .wincon_burstlen_shift = 11, +}; + +static struct decon_data exynos7870_decon_data = { + .vidw_buf_start_base = 0x880, + .shadowcon_win_protect_shift = 8, + .wincon_burstlen_shift = 10, +}; + struct decon_context { struct device *dev; struct drm_device *drm_dev; @@ -55,11 +73,19 @@ struct decon_context { wait_queue_head_t wait_vsync_queue; atomic_t wait_vsync_event; + const struct decon_data *data; struct drm_encoder *encoder; }; static const struct of_device_id decon_driver_dt_match[] = { - {.compatible = "samsung,exynos7-decon"}, + { + .compatible = "samsung,exynos7-decon", + .data = &exynos7_decon_data, + }, + { + .compatible = "samsung,exynos7870-decon", + .data = &exynos7870_decon_data, + }, {}, }; MODULE_DEVICE_TABLE(of, decon_driver_dt_match); @@ -92,8 +118,9 @@ static void decon_shadow_protect_win(struct decon_context *ctx, unsigned int win, bool protect) { u32 bits, val; + unsigned int shift = ctx->data->shadowcon_win_protect_shift; - bits = SHADOWCON_WINx_PROTECT(win); + bits = SHADOWCON_WINx_PROTECT(shift, win); val = readl(ctx->regs + SHADOWCON); if (protect) @@ -291,6 +318,7 @@ static void decon_win_set_pixfmt(struct decon_context *ctx, unsigned int win, { unsigned long val; int padding; + unsigned int shift = ctx->data->wincon_burstlen_shift; val = readl(ctx->regs + WINCON(win)); val &= ~WINCONx_BPPMODE_MASK; @@ -298,44 +326,44 @@ static void decon_win_set_pixfmt(struct decon_context *ctx, unsigned int win, switch (fb->format->format) { case DRM_FORMAT_RGB565: val |= WINCONx_BPPMODE_16BPP_565; - val |= WINCONx_BURSTLEN_16WORD; + val |= WINCONx_BURSTLEN_16WORD(shift); break; case DRM_FORMAT_XRGB8888: val |= WINCONx_BPPMODE_24BPP_xRGB; - val |= WINCONx_BURSTLEN_16WORD; + val |= WINCONx_BURSTLEN_16WORD(shift); break; case DRM_FORMAT_XBGR8888: val |= WINCONx_BPPMODE_24BPP_xBGR; - val |= WINCONx_BURSTLEN_16WORD; + val |= WINCONx_BURSTLEN_16WORD(shift); break; case DRM_FORMAT_RGBX8888: val |= WINCONx_BPPMODE_24BPP_RGBx; - val |= WINCONx_BURSTLEN_16WORD; + val |= WINCONx_BURSTLEN_16WORD(shift); break; case DRM_FORMAT_BGRX8888: val |= WINCONx_BPPMODE_24BPP_BGRx; - val |= WINCONx_BURSTLEN_16WORD; + val |= WINCONx_BURSTLEN_16WORD(shift); break; case DRM_FORMAT_ARGB8888: val |= WINCONx_BPPMODE_32BPP_ARGB | WINCONx_BLD_PIX | WINCONx_ALPHA_SEL; - val |= WINCONx_BURSTLEN_16WORD; + val |= WINCONx_BURSTLEN_16WORD(shift); break; case DRM_FORMAT_ABGR8888: val |= WINCONx_BPPMODE_32BPP_ABGR | WINCONx_BLD_PIX | WINCONx_ALPHA_SEL; - val |= WINCONx_BURSTLEN_16WORD; + val |= WINCONx_BURSTLEN_16WORD(shift); break; case DRM_FORMAT_RGBA8888: val |= WINCONx_BPPMODE_32BPP_RGBA | WINCONx_BLD_PIX | WINCONx_ALPHA_SEL; - val |= WINCONx_BURSTLEN_16WORD; + val |= WINCONx_BURSTLEN_16WORD(shift); break; case DRM_FORMAT_BGRA8888: default: val |= WINCONx_BPPMODE_32BPP_BGRA | WINCONx_BLD_PIX | WINCONx_ALPHA_SEL; - val |= WINCONx_BURSTLEN_16WORD; + val |= WINCONx_BURSTLEN_16WORD(shift); break; } @@ -351,8 +379,8 @@ static void decon_win_set_pixfmt(struct decon_context *ctx, unsigned int win, padding = (fb->pitches[0] / fb->format->cpp[0]) - fb->width; if (fb->width + padding < MIN_FB_WIDTH_FOR_16WORD_BURST) { - val &= ~WINCONx_BURSTLEN_MASK; - val |= WINCONx_BURSTLEN_8WORD; + val &= ~WINCONx_BURSTLEN_MASK(shift); + val |= WINCONx_BURSTLEN_8WORD(shift); } writel(val, ctx->regs + WINCON(win)); @@ -397,6 +425,7 @@ static void decon_update_plane(struct exynos_drm_crtc *crtc, unsigned int win = plane->index; unsigned int cpp = fb->format->cpp[0]; unsigned int pitch = fb->pitches[0]; + unsigned int vidw_addr0_base = ctx->data->vidw_buf_start_base; if (ctx->suspended) return; @@ -413,7 +442,7 @@ static void decon_update_plane(struct exynos_drm_crtc *crtc, /* buffer start address */ val = (unsigned long)exynos_drm_fb_dma_addr(fb, 0); - writel(val, ctx->regs + VIDW_BUF_START(win)); + writel(val, ctx->regs + VIDW_BUF_START(vidw_addr0_base, win)); padding = (pitch / cpp) - fb->width; @@ -695,6 +724,7 @@ static int decon_probe(struct platform_device *pdev) ctx->dev = dev; ctx->suspended = true; + ctx->data = of_device_get_match_data(dev); i80_if_timings = of_get_child_by_name(dev->of_node, "i80-if-timings"); if (i80_if_timings) diff --git a/drivers/gpu/drm/exynos/regs-decon7.h b/drivers/gpu/drm/exynos/regs-decon7.h index 5bc5f1db5196..216c106dac8f 100644 --- a/drivers/gpu/drm/exynos/regs-decon7.h +++ b/drivers/gpu/drm/exynos/regs-decon7.h @@ -48,7 +48,7 @@ /* SHADOWCON */ #define SHADOWCON 0x30 -#define SHADOWCON_WINx_PROTECT(_win) (1 << (10 + (_win))) +#define SHADOWCON_WINx_PROTECT(_shf, _win) (1 << ((_shf) + (_win))) /* WINCONx */ #define WINCON(_win) (0x50 + ((_win) * 4)) @@ -58,10 +58,9 @@ #define WINCONx_BUFSEL_SHIFT 28 #define WINCONx_TRIPLE_BUF_MODE (0x1 << 18) #define WINCONx_DOUBLE_BUF_MODE (0x0 << 18) -#define WINCONx_BURSTLEN_16WORD (0x0 << 11) -#define WINCONx_BURSTLEN_8WORD (0x1 << 11) -#define WINCONx_BURSTLEN_MASK (0x1 << 11) -#define WINCONx_BURSTLEN_SHIFT 11 +#define WINCONx_BURSTLEN_16WORD(_shf) (0x0 << (_shf)) +#define WINCONx_BURSTLEN_8WORD(_shf) (0x1 << (_shf)) +#define WINCONx_BURSTLEN_MASK(_shf) (0x1 << (_shf)) #define WINCONx_BLD_PLANE (0 << 8) #define WINCONx_BLD_PIX (1 << 8) #define WINCONx_ALPHA_MUL (1 << 7) @@ -89,9 +88,9 @@ #define VIDOSD_H(_x) (0x80 + ((_x) * 4)) /* Frame buffer start addresses: VIDWxxADD0n */ -#define VIDW_BUF_START(_win) (0x80 + ((_win) * 0x10)) -#define VIDW_BUF_START1(_win) (0x84 + ((_win) * 0x10)) -#define VIDW_BUF_START2(_win) (0x88 + ((_win) * 0x10)) +#define VIDW_BUF_START(_base, _win) ((_base) + ((_win) * 0x10)) +#define VIDW_BUF_START1(_base, _win) ((_base) + ((_win) * 0x10)) +#define VIDW_BUF_START2(_base, _win) ((_base) + ((_win) * 0x10)) #define VIDW_WHOLE_X(_win) (0x0130 + ((_win) * 8)) #define VIDW_WHOLE_Y(_win) (0x0134 + ((_win) * 8)) -- 2.51.0 From 3bca8bc1fc3a7c66a6a4a5e40b84f6cf128b70e7 Mon Sep 17 00:00:00 2001 From: Kaustabh Chakraborty Date: Wed, 6 Nov 2024 09:00:54 +0900 Subject: [PATCH 16/16] dt-bindings: display: samsung,exynos7-decon: add exynos7870 compatible Add the compatible string of Exynos7870 to the existing list. Signed-off-by: Kaustabh Chakraborty Acked-by: Krzysztof Kozlowski Signed-off-by: Inki Dae --- .../bindings/display/samsung/samsung,exynos7-decon.yaml | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/Documentation/devicetree/bindings/display/samsung/samsung,exynos7-decon.yaml b/Documentation/devicetree/bindings/display/samsung/samsung,exynos7-decon.yaml index 992c23ca7a4e..53916e4c95d8 100644 --- a/Documentation/devicetree/bindings/display/samsung/samsung,exynos7-decon.yaml +++ b/Documentation/devicetree/bindings/display/samsung/samsung,exynos7-decon.yaml @@ -19,7 +19,9 @@ description: | properties: compatible: - const: samsung,exynos7-decon + enum: + - samsung,exynos7-decon + - samsung,exynos7870-decon clocks: maxItems: 4 -- 2.51.0