Optimize dma_*_from_contiguous calls
Subsystems (2)
v3 cover Optimize dma_*_from_contiguous calls
[ Per discussion at v1, we decide to add two new functions and start
replacing callers one by one. For this series, it only touches the dma-direct part. And instead of merging two PATCHes, I still keep them separate so that we may easily revert PATCH-2 if anything bad happens as last time -- PATCH-1 is supposed to be a safe cleanup. ]
This series of patches try to optimize dma_*_from_contiguous calls:
PATCH-1 abstracts two new functions and applies to dma-direct.c file. PATCH-2 saves single pages and reduce fragmentations from CMA area.
Please check their commit messages for detail changelog.
Nicolin Chen (2):
dma-contiguous: Abstract dma_{alloc,free}_contiguous()
dma-contiguous: Use fallback alloc_pages for single pages
include/linux/dma-contiguous.h | 11 +++++++ kernel/dma/contiguous.c | 57 ++++++++++++++++++++++++++++++++++ kernel/dma/direct.c | 24 +++----------- 3 files changed, 72 insertions(+), 20 deletions(-)
Signature · 6 lines
2.17.1 _______________________________________________ iommu mailing list [email protected] https://lists.linuxfoundation.org/mailman/listinfo/iommu
PATCH v3 1/2 dma-contiguous: Abstract dma_{alloc,free}_contiguous()
Both dma_alloc_from_contiguous() and dma_release_from_contiguous() are very simply implemented, but requiring callers to pass certain parameters like count and align, and taking a boolean parameter to check __GFP_NOWARN in the allocation flags. So every function call duplicates similar work:
/* A piece of example */
unsigned long order = get_order(size);
size_t count = size >> PAGE_SHIFT;
page = dma_alloc_from_contiguous(dev, count, order, gfp & __GFP_NOWARN); [...]
dma_release_from_contiguous(dev, page, size >> PAGE_SHIFT);
Additionally, as CMA can be used only in the context which permits sleeping, most of callers do a gfpflags_allow_blocking() check and a corresponding fallback allocation of normal pages upon any false result:
/* A piece of example */
if (gfpflags_allow_blocking(flag))
page = dma_alloc_from_contiguous();
if (!page)
page = alloc_pages();
[...]
if (!dma_release_from_contiguous(dev, page, count))
__free_pages(page, get_order(size));So this patch simplifies those function calls by abstracting these operations into the two new functions: dma_{alloc,free}_contiguous.
As some callers of dma_{alloc,release}_from_contiguous() might be complicated, this patch just implements these two new functions to kernel/dma/direct.c only as an initial step.
---
Changelogv2->v3:
* Added missing "static inline" in header file to fix build error.v1->v2:
* Added new functions beside the old ones so we can replace callers one by one later.
* Applied new functions to dma/direct.c only, because it's the best example caller to apply and should be safe with the new functions.
include/linux/dma-contiguous.h | 11 ++++++++ kernel/dma/contiguous.c | 48 ++++++++++++++++++++++++++++++++++ kernel/dma/direct.c | 24 +++-------------- 3 files changed, 63 insertions(+), 20 deletions(-)
include/linux/dma-contiguous.h +11 -0
@@ -115,6 +115,8 @@ struct page *dma_alloc_from_contiguous(struct device *dev, size_t count, unsigned int order, bool no_warn);bool dma_release_from_contiguous (struct device *dev,struct page *pages,int count); +struct page *dma_alloc_contiguous (struct device *dev,size_t size,gfp_t gfp); +void dma_free_contiguous (struct device *dev,struct page *page,size_t size);#else
@@ -157,6 +159,15 @@ bool dma_release_from_contiguous(struct device *dev, struct page *pages,return false; } +static inline +struct page *dma_alloc_contiguous (struct device *dev,size_t size,gfp_t gfp) +{ +return NULL ; +} + +static inline +void dma_free_contiguous (struct device *dev,struct page *page,size_t size) { } +#endif #endif
kernel/dma/contiguous.c +48 -0
54 lines not on this page · raw message
kernel/dma/direct.c +4 -20
@@ -96,8 +96,6 @@ static bool dma_coherent_ok(struct device *dev, phys_addr_t phys, size_t size)struct page *__dma_direct_alloc_pages (struct device *dev,size_t size,dma_addr_t *dma_handle,gfp_t gfp,unsigned long attrs) { -unsigned int count =PAGE_ALIGN (size) >>PAGE_SHIFT ; -int page_order =get_order (size);struct page *page =NULL ;u64 phys_mask;
@@ -109,20 +107,9 @@ struct page *__dma_direct_alloc_pages(struct device *dev, size_t size, gfp |=__dma_direct_optimal_gfp_mask (dev, dev->coherent_dma_mask, &phys_mask); again: -/* CMA can be used only in the context which permits sleeping */ -if (gfpflags_allow_blocking (gfp)) { - page =dma_alloc_from_contiguous (dev, count, page_order, - gfp & __GFP_NOWARN); -if (page && !dma_coherent_ok (dev,page_to_phys (page), size)) { -dma_release_from_contiguous (dev, page, count); - page =NULL ; - } - } -if (!page) - page =alloc_pages_node (dev_to_node (dev), gfp, page_order); - + page =dma_alloc_contiguous (dev, size, gfp);if (page && !dma_coherent_ok (dev,page_to_phys (page), size)) { -__free_pages (page, page_order); +dma_free_contiguous (dev, page, size); page =NULL ; if (IS_ENABLED(CONFIG_ZONE_DMA32) &&
@@ -154,7 +141,7 @@ void *dma_direct_alloc_pages(struct device *dev, size_t size,if (PageHighMem (page)) {/* * Depending on the cma= arguments and per-arch setup -* dma_alloc_from_contiguous could return highmem pages. +* dma_alloc_contiguous could return highmem pages. * Without remapping there is no way to return them here, * so log an error and fail. */
@@ -176,10 +163,7 @@ void *dma_direct_alloc_pages(struct device *dev, size_t size,void __dma_direct_free_pages (struct device *dev,size_t size,struct page *page) { -unsigned int count =PAGE_ALIGN (size) >>PAGE_SHIFT ; - -if (!dma_release_from_contiguous (dev, page, count)) -__free_pages (page,get_order (size)); +dma_free_contiguous (dev, page, size); } void dma_direct_free_pages(struct device *dev, size_t size, void *cpu_addr,
Signature · 1 line
2.17.1
Re: [PATCH v3 1/2] dma-contiguous: Abstract dma_{alloc,free}_contiguous()
Hi Nicolin,
Quoted 28 lines · Nicolin Chen
Both
dma_alloc_from_contiguous()anddma_release_from_contiguous()are very simply implemented, but requiring callers to pass certain parameters like count and align, and taking a boolean parameter to check__GFP_NOWARNin the allocation flags. So every function call duplicates similar work:
/* A piece of example */
unsigned long order =get_order(size);
size_tcount = size >>PAGE_SHIFT;
page =dma_alloc_from_contiguous(dev, count, order, gfp & __GFP_NOWARN); [...]
dma_release_from_contiguous(dev, page, size >> PAGE_SHIFT);Additionally, as CMA can be used only in the context which permits sleeping, most of callers do a
gfpflags_allow_blocking()check and a corresponding fallback allocation of normal pages upon any false result:
/* A piece of example */
if (gfpflags_allow_blocking(flag))page = dma_alloc_from_contiguous(); if (!page) page = alloc_pages(); [...] if (!dma_release_from_contiguous(dev, page, count)) __free_pages(page, get_order(size));So this patch simplifies those function calls by abstracting these operations into the two new functions: dma_{alloc,free}_contiguous.
Quoted 7 lines
As some callers of dma_{alloc,release}_from_contiguous() might be complicated, this patch just implements these two new functions to
kernel/dma/direct.conly as an initial step.Suggested-by Christoph Hellwig <[email protected]>Signed-off-by Nicolin Chen <[email protected]>---
This commit is causing boot failures in QEMU on x86_64 defconfig:
https://travis-ci.com/ClangBuiltLinux/continuous-integration/jobs/203825363
Attached is a bisect log and a boot log with GCC (just to show it is not a compiler thing).
My QEMU command line is:
qemu-system-x86_64 -m 512m \
-drive file=images/x86_64/rootfs.ext4,format=raw,if=ide \
-append 'console=ttyS0 root=/dev/sda' \
-nographic \
-kernel arch/x86_64/boot/bzImageand the rootfs is available here:
https://github.com/ClangBuiltLinux/continuous-integration/raw/master/images/x86_64/rootfs.ext4
I haven't seen a report on this yet so apologize if there is already a fix in the works. Let me know if you need anythnig from me.
Cheers,
Nathan
git bisect start
# good: [9fb67d643f6f1892a08ee3a04ea54022d1060bb0] Merge tag 'pinctrl-v5.2-2' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl
git bisect good 9fb67d643f6f1892a08ee3a04ea54022d1060bb0
# bad: [9a15d2e3fd03e38a6ee7d7bc34d28bb7340f05f2] Add linux-next specific files for 20190529
git bisect bad 9a15d2e3fd03e38a6ee7d7bc34d28bb7340f05f2
# bad: [eb756b9cce06b6c30b478a7ead67ddc0aa52b421] Merge remote-tracking branch 'crypto/master'
git bisect bad eb756b9cce06b6c30b478a7ead67ddc0aa52b421
# bad: [ffd5fc17ee0cd5f90258a8d4e0b87af913943f72] Merge remote-tracking branch 'xtensa/xtensa-for-next'
git bisect bad ffd5fc17ee0cd5f90258a8d4e0b87af913943f72
# bad: [df846fa0031f0e11dbfd5dd7c959cb45b3c0b3e2] Merge remote-tracking branch 'actions/for-next'
git bisect bad df846fa0031f0e11dbfd5dd7c959cb45b3c0b3e2
# good: [673e28ea2b579adcec369cd7f7295142a6b7e017] Merge remote-tracking branch 'usb.current/usb-linus'
git bisect good 673e28ea2b579adcec369cd7f7295142a6b7e017
# good: [d4f1f6efe84b20ff5f3a4874c580c2bee6cba68a] Merge remote-tracking branch 'kvms390-fixes/master'
git bisect good d4f1f6efe84b20ff5f3a4874c580c2bee6cba68a
# bad: [d0756fb0025937a5ed866d27f4452c66b066c089] Merge remote-tracking branch 'dma-mapping/for-next'
git bisect bad d0756fb0025937a5ed866d27f4452c66b066c089
# good: [49af21d0007414838a27eb4ade21277350e8ef1d] Merge remote-tracking branch 'drm-misc-fixes/for-linux-next-fixes'
git bisect good 49af21d0007414838a27eb4ade21277350e8ef1d
# good: [8680aa5a58abfe6087a3d8248c02232d3e05dc80] iommu/dma: Don't remap CMA unnecessarily
git bisect good 8680aa5a58abfe6087a3d8248c02232d3e05dc80
# good: [efd9f10b70689fdeacadc95b6e0ea6dc311fa64f] iommu/dma: Refactor iommu_dma_mmap
git bisect good efd9f10b70689fdeacadc95b6e0ea6dc311fa64f
# good: [b5f75a3639ff3b547e4eee7671e4321a429747a6] arm64: switch copyright boilerplace to SPDX in dma-mapping.c
git bisect good b5f75a3639ff3b547e4eee7671e4321a429747a6
# bad: [fdaeec198ada8c48bff03c85fab542e5b241f5bc] dma-contiguous: add dma_{alloc,free}_contiguous() helpers
git bisect bad fdaeec198ada8c48bff03c85fab542e5b241f5bc
# good: [a84cc69eb53715d37242a21ada398b0d8cd316fc] arm64: trim includes in dma-mapping.c
git bisect good a84cc69eb53715d37242a21ada398b0d8cd316fc
# first bad commit: [fdaeec198ada8c48bff03c85fab542e5b241f5bc] dma-contiguous: add dma_{alloc,free}_contiguous() helpersc[?7l[2J[0mSeaBIOS (version 1.12.0-20181126_142135-anatol)
iPXE (http://ipxe.org) 00:03.0 C980 PCI2.10 PnP PMM+1FF92110+1FEF2110 C980 Press Ctrl-B to configure iPXE (PCI 00:03.0)...
Booting from ROM..c[?7l[2J[0m. [ 0.000000] Linux version 5.2.0-rc2-next-20190529 (nathan@archlinux-epyc) (gcc version 8.3.0 (GCC)) #1 SMP Wed May 29 11:26:28 MST 2019
[ 0.000000] Command line: root=/dev/sda console=ttyS0 [ 0.000000] x86/fpu: Supporting XSAVE feature 0x001: 'x87 floating point registers' [ 0.000000] x86/fpu: Supporting XSAVE feature 0x002: 'SSE registers' [ 0.000000] x86/fpu: Supporting XSAVE feature 0x004: 'AVX registers' [ 0.000000] x86/fpu: xstate_offset[2]: 576, xstate_sizes[2]: 256 [ 0.000000] x86/fpu: Enabled xstate features 0x7, context size is 832 bytes, using 'standard' format. [ 0.000000] BIOS-provided physical RAM map: [ 0.000000] BIOS-e820: [mem 0x0000000000000000-0x000000000009fbff] usable [ 0.000000] BIOS-e820: [mem 0x000000000009fc00-0x000000000009ffff] reserved [ 0.000000] BIOS-e820: [mem 0x00000000000f0000-0x00000000000fffff] reserved [ 0.000000] BIOS-e820: [mem 0x0000000000100000-0x000000001ffdffff] usable [ 0.000000] BIOS-e820: [mem 0x000000001ffe0000-0x000000001fffffff] reserved [ 0.000000] BIOS-e820: [mem 0x00000000feffc000-0x00000000feffffff] reserved [ 0.000000] BIOS-e820: [mem 0x00000000fffc0000-0x00000000ffffffff] reserved [ 0.000000] NX (Execute Disable) protection: active [ 0.000000] SMBIOS 2.8 present. [ 0.000000] DMI: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.12.0-20181126_142135-anatol 04/01/2014 [ 0.000000] tsc: Fast TSC calibration using PIT [ 0.000000] tsc: Detected 1999.996 MHz processor [ 0.001281] last_pfn = 0x1ffe0 max_arch_pfn = 0x400000000 [ 0.001327] x86/PAT: Configuration [0-7]: WB WC UC- UC WB WP UC- WT [ 0.003614] found SMP MP-table at [mem 0x000f5cc0-0x000f5ccf] [ 0.003654] check: Scanning 1 areas for low memory corruption [ 0.003674] Using GB pages for direct mapping [ 0.003756] ACPI: Early table checksum verification disabled [ 0.003788] ACPI: RSDP 0x00000000000F5AF0 000014 (v00 BOCHS ) [ 0.003793] ACPI: RSDT 0x000000001FFE156F 000030 (v01 BOCHS BXPCRSDT 00000001 BXPC 00000001) [ 0.003798] ACPI: FACP 0x000000001FFE144B 000074 (v01 BOCHS BXPCFACP 00000001 BXPC 00000001) [ 0.003802] ACPI: DSDT 0x000000001FFE0040 00140B (v01 BOCHS BXPCDSDT 00000001 BXPC 00000001) [ 0.003804] ACPI: FACS 0x000000001FFE0000 000040 [ 0.003806] ACPI: APIC 0x000000001FFE14BF 000078 (v01 BOCHS BXPCAPIC 00000001 BXPC 00000001) [ 0.003808] ACPI: HPET 0x000000001FFE1537 000038 (v01 BOCHS BXPCHPET 00000001 BXPC 00000001) [ 0.004087] No NUMA configuration found [ 0.004088] Faking a node at [mem 0x0000000000000000-0x000000001ffdffff] [ 0.004091] NODE_DATA(0) allocated [mem 0x1ffdc000-0x1ffdffff] [ 0.004328] Zone ranges: [ 0.004330] DMA [mem 0x0000000000001000-0x0000000000ffffff] [ 0.004331] DMA32 [mem 0x0000000001000000-0x000000001ffdffff] [ 0.004332] Normal empty [ 0.004332] Movable zone start for each node [ 0.004333] Early memory node ranges [ 0.004334] node 0: [mem 0x0000000000001000-0x000000000009efff] [ 0.004335] node 0: [mem 0x0000000000100000-0x000000001ffdffff] [ 0.004612] Zeroed struct page in unavailable ranges: 98 pages [ 0.004613] Initmem setup node 0 [mem 0x0000000000001000-0x000000001ffdffff] [ 0.006924] ACPI: PM-Timer IO Port: 0x608 [ 0.006930] ACPI: LAPIC_NMI (acpi_id[0xff] dfl dfl lint[0x1]) [ 0.006958] IOAPIC[0]: apic_id 0, version 17, address 0xfec00000, GSI 0-23 [ 0.006960] ACPI: INT_SRC_OVR (bus 0 bus_irq 0 global_irq 2 dfl dfl) [ 0.006961] ACPI: INT_SRC_OVR (bus 0 bus_irq 5 global_irq 5 high level) [ 0.006962] ACPI: INT_SRC_OVR (bus 0 bus_irq 9 global_irq 9 high level) [ 0.006963] ACPI: INT_SRC_OVR (bus 0 bus_irq 10 global_irq 10 high level) [ 0.006964] ACPI: INT_SRC_OVR (bus 0 bus_irq 11 global_irq 11 high level) [ 0.006967] Using ACPI (MADT) for SMP configuration information [ 0.006968] ACPI: HPET id: 0x8086a201 base: 0xfed00000 [ 0.006974] smpboot: Allowing 1 CPUs, 0 hotplug CPUs [ 0.006982] PM: Registered nosave memory: [mem 0x00000000-0x00000fff] [ 0.006983] PM: Registered nosave memory: [mem 0x0009f000-0x0009ffff] [ 0.006983] PM: Registered nosave memory: [mem 0x000a0000-0x000effff] [ 0.006984] PM: Registered nosave memory: [mem 0x000f0000-0x000fffff] [ 0.006986] [mem 0x20000000-0xfeffbfff] available for PCI devices [ 0.006989] clocksource: refined-jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 1910969940391419 ns [ 0.100715] setup_percpu: NR_CPUS:64 nr_cpumask_bits:64 nr_cpu_ids:1 nr_node_ids:1 [ 0.101067] percpu: Embedded 51 pages/cpu s170968 r8192 d29736 u2097152 [ 0.101083] node[0] zonelist: 0:DMA32 0:DMA [ 0.101086] Built 1 zonelists, mobility grouping on. Total pages: 128873 [ 0.101087] Policy zone: DMA32 [ 0.101089] Kernel command line: root=/dev/sda console=ttyS0 [ 0.102238] Memory: 486332K/523768K available (14340K kernel code, 1311K rwdata, 3128K rodata, 1308K init, 1276K bss, 37436K reserved, 0K cma-reserved) [ 0.102591] SLUB: HWalign=64, Order=0-3, MinObjects=0, CPUs=1, Nodes=1 [ 0.102930] rcu: Hierarchical RCU implementation. [ 0.102931] rcu: RCU event tracing is enabled. [ 0.102932] rcu: RCU restricting CPUs from NR_CPUS=64 to nr_cpu_ids=1. [ 0.102933] rcu: RCU calculated value of scheduler-enlistment delay is 100 jiffies. [ 0.102934] rcu: Adjusting geometry for rcu_fanout_leaf=16, nr_cpu_ids=1 [ 0.103054] NR_IRQS: 4352, nr_irqs: 256, preallocated irqs: 16 [ 0.103367] random: get_random_bytes called from start_kernel+0x2c4/0x498 with crng_init=0 [ 0.110098] Console: colour VGA+ 80x25 [ 0.176180] printk: console [ttyS0] enabled [ 0.176728] ACPI: Core revision 20190509 [ 0.177463] clocksource: hpet: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 19112604467 ns [ 0.178714] APIC: Switch to symmetric I/O mode setup [ 0.180399] ..TIMER: vector=0x30 apic1=0 pin1=2 apic2=-1 pin2=-1 [ 0.185697] clocksource: tsc-early: mask: 0xffffffffffffffff max_cycles: 0x39a8554e05d, max_idle_ns: 881590540420 ns [ 0.187046] Calibrating delay loop (skipped), value calculated using timer frequency.. 3999.99 BogoMIPS (lpj=1999996) [ 0.188046] pid_max: default: 32768 minimum: 301 [ 0.189060] LSM: Security Framework initializing [ 0.189653] SELinux: Initializing. [ 0.190054] *** VALIDATE SELinux *** [ 0.190556] Dentry cache hash table entries: 65536 (order: 7, 524288 bytes) [ 0.191100] Inode-cache hash table entries: 32768 (order: 6, 262144 bytes) [ 0.192058] Mount-cache hash table entries: 1024 (order: 1, 8192 bytes) [ 0.193047] Mountpoint-cache hash table entries: 1024 (order: 1, 8192 bytes) [ 0.194137] *** VALIDATE proc *** [ 0.195066] *** VALIDATE cgroup1 *** [ 0.195530] *** VALIDATE cgroup2 *** [ 0.196165] Last level iTLB entries: 4KB 512, 2MB 255, 4MB 127 [ 0.197046] Last level dTLB entries: 4KB 512, 2MB 255, 4MB 127, 1GB 0 [ 0.197853] Spectre V2 : Mitigation: Full AMD retpoline [ 0.198045] Spectre V2 : Spectre v2 / SpectreRSB mitigation: Filling RSB on context switch [ 0.199046] Spectre V2 : mitigation: Enabling conditional Indirect Branch Prediction Barrier [ 0.200046] Speculative Store Bypass: Mitigation: Speculative Store Bypass disabled via prctl and seccomp [ 0.205525] Freeing SMP alternatives memory: 40K [ 0.206148] smpboot: CPU0: AMD EPYC 7401P 24-Core Processor (family: 0x17, model: 0x1, stepping: 0x2) [ 0.207115] Performance Events: Fam17h core perfctr, AMD PMU driver. [ 0.207932] ... version: 0 [ 0.208047] ... bit width: 48 [ 0.208573] ... generic registers: 6 [ 0.209047] ... value mask: 0000ffffffffffff [ 0.209714] ... max period: 00007fffffffffff [ 0.210046] ... fixed-purpose events: 0 [ 0.210556] ... event mask: 000000000000003f [ 0.211077] rcu: Hierarchical SRCU implementation. [ 0.211716] Decoding supported only on Scalable MCA processors. [ 0.212065] smp: Bringing up secondary CPUs ... [ 0.212640] smp: Brought up 1 node, 1 CPU [ 0.213047] smpboot: Max logical packages: 1 [ 0.213588] smpboot: Total of 1 processors activated (3999.99 BogoMIPS) [ 0.214176] devtmpfs: initialized [ 0.214753] clocksource: jiffies: mask: 0xffffffff max_cycles: 0xffffffff, max_idle_ns: 1911260446275000 ns [ 0.215050] futex hash table entries: 256 (order: 2, 16384 bytes) [ 0.215897] PM: RTC time: 18:27:49, date: 2019-05-29 [ 0.216386] NET: Registered protocol family 16 [ 0.217012] audit: initializing netlink subsys (disabled) [ 0.217253] cpuidle: using governor menu [ 0.218179] ACPI: bus type PCI registered [ 0.218756] PCI: Using configuration type 1 for base access [ 0.219047] PCI: Using configuration type 1 for extended access [ 0.220053] audit: type=2000 audit(1559154469.038:1): state=initialized audit_enabled=0 res=1 [ 0.222328] HugeTLB registered 2.00 MiB page size, pre-allocated 0 pages [ 0.223111] cryptomgr_test (20) used greatest stack depth: 15520 bytes left [ 0.224013] kworker/u2:0 (22) used greatest stack depth: 14664 bytes left [ 0.224111] kworker/u2:0 (26) used greatest stack depth: 14216 bytes left [ 0.225711] ACPI: Added _OSI(Module Device) [ 0.226051] ACPI: Added _OSI(Processor Device) [ 0.226618] ACPI: Added _OSI(3.0 _SCP Extensions) [ 0.227047] ACPI: Added _OSI(Processor Aggregator Device) [ 0.227737] ACPI: Added _OSI(Linux-Dell-Video) [ 0.228071] ACPI: Added _OSI(Linux-Lenovo-NV-HDMI-Audio) [ 0.228741] ACPI: Added _OSI(Linux-HPI-Hybrid-Graphics) [ 0.229545] ACPI: 1 ACPI AML tables successfully acquired and loaded [ 0.231171] ACPI: Interpreter enabled [ 0.231666] ACPI: (supports S0 S3 S4 S5) [ 0.232048] ACPI: Using IOAPIC for interrupt routing [ 0.232683] PCI: Using host bridge windows from ACPI; if necessary, use "pci=nocrs" and report a bug [ 0.233119] ACPI: Enabled 2 GPEs in block 00 to 0F [ 0.235370] ACPI: PCI Root Bridge [PCI0] (domain 0000 [bus 00-ff]) [ 0.236052] acpi PNP0A03:00: _OSC: OS supports [ExtendedConfig ASPM ClockPM Segments MSI HPX-Type3] [ 0.237078] PCI host bridge to bus 0000:00 [ 0.237602] pci_bus 0000:00: root bus resource [io 0x0000-0x0cf7 window] [ 0.238048] pci_bus 0000:00: root bus resource [io 0x0d00-0xffff window] [ 0.238942] pci_bus 0000:00: root bus resource [mem 0x000a0000-0x000bffff window] [ 0.239048] pci_bus 0000:00: root bus resource [mem 0x20000000-0xfebfffff window] [ 0.240014] pci_bus 0000:00: root bus resource [mem 0x100000000-0x17fffffff window] [ 0.240047] pci_bus 0000:00: root bus resource [bus 00-ff] [ 0.240789] pci 0000:00:00.0: [8086:1237] type 00 class 0x060000 [ 0.241446] pci 0000:00:01.0: [8086:7000] type 00 class 0x060100 [ 0.242459] pci 0000:00:01.1: [8086:7010] type 00 class 0x010180 [ 0.244902] pci 0000:00:01.1: reg 0x20: [io 0xc040-0xc04f] [ 0.245838] pci 0000:00:01.1: legacy IDE quirk: reg 0x10: [io 0x01f0-0x01f7] [ 0.246049] pci 0000:00:01.1: legacy IDE quirk: reg 0x14: [io 0x03f6] [ 0.246897] pci 0000:00:01.1: legacy IDE quirk: reg 0x18: [io 0x0170-0x0177] [ 0.247047] pci 0000:00:01.1: legacy IDE quirk: reg 0x1c: [io 0x0376] [ 0.248021] pci 0000:00:01.3: [8086:7113] type 00 class 0x068000 [ 0.248360] pci 0000:00:01.3: quirk: [io 0x0600-0x063f] claimed by PIIX4 ACPI [ 0.249053] pci 0000:00:01.3: quirk: [io 0x0700-0x070f] claimed by PIIX4 SMB [ 0.250137] pci 0000:00:02.0: [1234:1111] type 00 class 0x030000 [ 0.252059] pci 0000:00:02.0: reg 0x10: [mem 0xfd000000-0xfdffffff pref] [ 0.256072] pci 0000:00:02.0: reg 0x18: [mem 0xfebf0000-0xfebf0fff] [ 0.263058] pci 0000:00:02.0: reg 0x30: [mem 0xfebe0000-0xfebeffff pref] [ 0.264378] pci 0000:00:03.0: [8086:100e] type 00 class 0x020000 [ 0.266050] pci 0000:00:03.0: reg 0x10: [mem 0xfebc0000-0xfebdffff] [ 0.268049] pci 0000:00:03.0: reg 0x14: [io 0xc000-0xc03f] [ 0.274058] pci 0000:00:03.0: reg 0x30: [mem 0xfeb80000-0xfebbffff pref] [ 0.276310] ACPI: PCI Interrupt Link [LNKA] (IRQs 5 *10 11) [ 0.277301] ACPI: PCI Interrupt Link [LNKB] (IRQs 5 *10 11) [ 0.278248] ACPI: PCI Interrupt Link [LNKC] (IRQs 5 10 *11) [ 0.279247] ACPI: PCI Interrupt Link [LNKD] (IRQs 5 10 *11) [ 0.280166] ACPI: PCI Interrupt Link [LNKS] (IRQs *9) [ 0.281410] pci 0000:00:02.0: vgaarb: setting as boot VGA device [ 0.282044] pci 0000:00:02.0: vgaarb: VGA device added: decodes=io+mem,owns=io+mem,locks=none [ 0.282056] pci 0000:00:02.0: vgaarb: bridge control possible [ 0.283054] vgaarb: loaded [ 0.284164] SCSI subsystem initialized [ 0.285174] ACPI: bus type USB registered [ 0.286088] usbcore: registered new interface driver usbfs [ 0.287074] usbcore: registered new interface driver hub [ 0.288087] usbcore: registered new device driver usb [ 0.289088] pps_core: LinuxPPS API ver. 1 registered [ 0.290054] pps_core: Software ver. 5.3.6 - Copyright 2005-2007 Rodolfo Giometti <[email protected]> [ 0.291064] PTP clock support registered [ 0.292127] EDAC MC: Ver: 3.0.0 [ 0.293326] Advanced Linux Sound Architecture Driver Initialized. [ 0.294072] PCI: Using ACPI for IRQ routing [ 0.295369] NetLabel: Initializing [ 0.296052] NetLabel: domain hash size = 128 [ 0.297050] NetLabel: protocols = UNLABELED CIPSOv4 CALIPSO [ 0.298079] NetLabel: unlabeled traffic allowed by default [ 0.299200] hpet0: at MMIO 0xfed00000, IRQs 2, 8, 0 [ 0.300053] hpet0: 3 comparators, 64-bit 100.000000 MHz counter [ 0.306070] clocksource: Switched to clocksource tsc-early [ 0.379516] VFS: Disk quotas dquot_6.6.0 [ 0.380072] VFS: Dquot-cache hash table entries: 512 (order 0, 4096 bytes) [ 0.380975] *** VALIDATE hugetlbfs *** [ 0.381470] pnp: PnP ACPI init [ 0.382179] pnp: PnP ACPI: found 6 devices [ 0.388273] clocksource: acpi_pm: mask: 0xffffff max_cycles: 0xffffff, max_idle_ns: 2085701024 ns [ 0.389427] pci_bus 0000:00: resource 4 [io 0x0000-0x0cf7 window] [ 0.390251] pci_bus 0000:00: resource 5 [io 0x0d00-0xffff window] [ 0.391054] pci_bus 0000:00: resource 6 [mem 0x000a0000-0x000bffff window] [ 0.391914] pci_bus 0000:00: resource 7 [mem 0x20000000-0xfebfffff window] [ 0.392777] pci_bus 0000:00: resource 8 [mem 0x100000000-0x17fffffff window] [ 0.393680] NET: Registered protocol family 2 [ 0.394330] tcp_listen_portaddr_hash hash table entries: 256 (order: 0, 4096 bytes) [ 0.395302] TCP established hash table entries: 4096 (order: 3, 32768 bytes) [ 0.396246] TCP bind hash table entries: 4096 (order: 4, 65536 bytes) [ 0.397089] TCP: Hash tables configured (established 4096 bind 4096) [ 0.397884] UDP hash table entries: 256 (order: 1, 8192 bytes) [ 0.398624] UDP-Lite hash table entries: 256 (order: 1, 8192 bytes) [ 0.399421] NET: Registered protocol family 1 [ 0.400031] RPC: Registered named UNIX socket transport module. [ 0.400773] RPC: Registered udp transport module. [ 0.401367] RPC: Registered tcp transport module. [ 0.401958] RPC: Registered tcp NFSv4.1 backchannel transport module. [ 0.402824] pci 0000:00:01.0: PIIX3: Enabling Passive Release [ 0.403549] pci 0000:00:00.0: Limiting direct PCI/PCI transfers [ 0.404288] pci 0000:00:01.0: Activating ISA DMA hang workarounds [ 0.405080] pci 0000:00:02.0: Video device with shadowed ROM at [mem 0x000c0000-0x000dffff] [ 0.406129] PCI: CLS 0 bytes, default 64 [ 0.406744] check: Scanning for low memory corruption every 60 seconds [ 0.407727] Initialise system trusted keyrings [ 0.408320] workingset: timestamp_bits=56 max_order=17 bucket_order=0 [ 0.410320] NFS: Registering the id_resolver key type [ 0.410953] Key type id_resolver registered [ 0.411481] Key type id_legacy registered [ 0.412107] Key type asymmetric registered [ 0.412623] Asymmetric key parser 'x509' registered [ 0.413235] Block layer SCSI generic (bsg) driver version 0.4 loaded (major 251) [ 0.414152] io scheduler mq-deadline registered [ 0.414715] io scheduler kyber registered [ 0.415305] input: Power Button as /devices/LNXSYSTM:00/LNXPWRBN:00/input/input0 [ 0.416234] ACPI: Power Button [PWRF] [ 0.416819] Serial: 8250/16550 driver, 4 ports, IRQ sharing enabled [ 0.440750] 00:05: ttyS0 at I/O 0x3f8 (irq = 4, base_baud = 115200) is a 16550A [ 0.441978] Non-volatile memory driver v1.3 [ 0.442575] Linux agpgart interface v0.103 [ 0.444240] loop: module loaded [ 0.444948] ata_piix 0000:00:01.1: failed to start port 0 (errno=-12) [ 0.445938] ata_piix: probe of 0000:00:01.1 failed with error -12 [ 0.446748] e100: Intel(R) PRO/100 Network Driver, 3.5.24-k2-NAPI [ 0.447504] e100: Copyright(c) 1999-2006 Intel Corporation [ 0.448196] e1000: Intel(R) PRO/1000 Network Driver - version 7.3.21-k8-NAPI [ 0.449116] e1000: Copyright (c) 1999-2006 Intel Corporation. [ 0.462026] PCI Interrupt Link [LNKC] enabled at IRQ 11 [ 0.764450] e1000 0000:00:03.0 eth0: (PCI:33MHz:32-bit) 52:54:00:12:34:56 [ 0.765305] e1000 0000:00:03.0 eth0: Intel(R) PRO/1000 Network Connection [ 0.766192] e1000e: Intel(R) PRO/1000 Network Driver - 3.2.6-k [ 0.766955] e1000e: Copyright(c) 1999 - 2015 Intel Corporation. [ 0.767732] sky2: driver version 1.30 [ 0.768275] ehci_hcd: USB 2.0 'Enhanced' Host Controller (EHCI) Driver [ 0.769128] ehci-pci: EHCI PCI platform driver [ 0.769710] ohci_hcd: USB 1.1 'Open' Host Controller (OHCI) Driver [ 0.770503] ohci-pci: OHCI PCI platform driver [ 0.771081] uhci_hcd: USB Universal Host Controller Interface driver [ 0.771910] usbcore: registered new interface driver usblp [ 0.772628] usbcore: registered new interface driver usb-storage [ 0.773424] i8042: PNP: PS/2 Controller [PNP0303:KBD,PNP0f13:MOU] at 0x60,0x64 irq 1,12 [ 0.775080] serio: i8042 KBD port at 0x60,0x64 irq 1 [ 0.775729] serio: i8042 AUX port at 0x60,0x64 irq 12 [ 0.776961] input: AT Translated Set 2 keyboard as /devices/platform/i8042/serio0/input/input1 [ 0.779751] rtc_cmos 00:00: RTC can wake from S4 [ 0.780682] rtc_cmos 00:00: registered as rtc0 [ 0.781298] rtc_cmos 00:00: alarms up to one day, y3k, 114 bytes nvram, hpet irqs [ 0.782350] device-mapper: ioctl: 4.40.0-ioctl (2019-01-18) initialised: [email protected] [ 0.783478] hidraw: raw HID events driver (C) Jiri Kosina [ 0.784247] usbcore: registered new interface driver usbhid [ 0.784940] usbhid: USB HID core driver [ 0.785877] Initializing XFRM netlink socket [ 0.786497] NET: Registered protocol family 10 [ 0.787219] Segment Routing with IPv6 [ 0.787713] sit: IPv6, IPv4 and MPLS over IPv4 tunneling driver [ 0.788537] NET: Registered protocol family 17 [ 0.789122] Key type dns_resolver registered [ 0.789708] mce: Using 10 MCE banks [ 0.790165] sched_clock: Marking stable (714371390, 75773427)->(925568491, -135423674) [ 0.791221] registered taskstats version 1 [ 0.791733] Loading compiled-in X.509 certificates [ 0.792509] PM: Magic number: 3:11:495 [ 0.793015] printk: console [netcon0] enabled [ 0.793565] netconsole: network logging started [ 0.794166] cfg80211: Loading compiled-in X.509 certificates for regulatory database [ 0.795403] cfg80211: Loaded X.509 cert 'sforshee: 00b28ddf47aef9cea7' [ 0.796230] ALSA device list: [ 0.796679] No soundcards found. [ 0.797386] platform regulatory.0: Direct firmware load for regulatory.db failed with error -2 [ 0.798530] cfg80211: failed to load regulatory.db [ 1.380351] input: ImExPS/2 Generic Explorer Mouse as /devices/platform/i8042/serio1/input/input3 [ 1.383202] md: Waiting for all devices to be available before autodetect [ 1.385280] md: If you don't use raid, use raid=noautodetect [ 1.387215] md: Autodetecting RAID arrays. [ 1.388498] md: autorun ... [ 1.389381] md: ... autorun DONE. [ 1.390461] VFS: Cannot open root device "sda" or unknown-block(0,0): error -6 [ 1.392649] Please append a correct "root=" boot option; here are the available partitions: [ 1.395141] Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(0,0) [ 1.397600] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.2.0-rc2-next-20190529 #1 [ 1.399793] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.12.0-20181126_142135-anatol 04/01/2014 [ 1.402816] Call Trace: [ 1.403577] dump_stack+0x46/0x60 [ 1.404586] panic+0xf6/0x2b7 [ 1.405512] mount_block_root+0x191/0x23b [ 1.406724] ? do_early_param+0x89/0x89 [ 1.407881] mount_root+0x10a/0x128 [ 1.408952] prepare_namespace+0x130/0x166 [ 1.410195] kernel_init_freeable+0x1df/0x1ea [ 1.411501] ? rest_init+0x9a/0x9a [ 1.412533] kernel_init+0x5/0xf6 [ 1.413536] ret_from_fork+0x22/0x40 [ 1.417612] Kernel Offset: 0x800000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff) [ 1.420749] ---[ end Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(0,0) ]---
Re: [PATCH v3 1/2] dma-contiguous: Abstract dma_{alloc,free}_contiguous()
Hi Nathan,
Quoted 15 lines · Nathan Chancellor
This commit is causing boot failures in QEMU on
x86_64defconfig:https://travis-ci.com/ClangBuiltLinux/continuous-integration/jobs/203825363
Attached is a bisect log and a boot log with GCC (just to show it is not a compiler thing).
My QEMU command line is:
qemu-system-x86_64 -m 512m \ -drive file=images/x86_64/rootfs.ext4,format=raw,if=ide \ -append 'console=ttyS0 root=/dev/sda' \ -nographic \ -kernel arch/x86_64/boot/bzImage
Quoted 3 lines
and the rootfs is available here:
https://github.com/ClangBuiltLinux/continuous-integration/raw/master/images/x86_64/rootfs.ext4
Thanks for reporting the bug.
I am able to repro the issue with the given command and rootfs. The problem is that x86_64 has CONFIG_DMA_CMA=n so the helper function is blank on x86_64 while dma-direct should be platform independent.
A simple fix is to add alloc_pages_node() for !CONFIG_DMA_CMA. I'll submit a fix soon -- need to figure out a good way though. It seems that adding the fallback to the !CONFIG_DMA_CMA version would cause some recipe errors when building the kernel...
Re: [PATCH v3 1/2] dma-contiguous: Abstract dma_{alloc,free}_contiguous()
In reply to Nicolin ChenI can't find a way to forward-redirect from Gmail, so I'm Ccing Dafna who found a regression caused by this commit. Dafna, can you give all the details, including the log and how you are reproducing it?
Quoted 198 lines · Nicolin Chen
Both
dma_alloc_from_contiguous()anddma_release_from_contiguous()are very simply implemented, but requiring callers to pass certain parameters like count and align, and taking a boolean parameter to check__GFP_NOWARNin the allocation flags. So every function call duplicates similar work:
/* A piece of example */
unsigned long order =get_order(size);
size_tcount = size >>PAGE_SHIFT;
page =dma_alloc_from_contiguous(dev, count, order, gfp & __GFP_NOWARN); [...]
dma_release_from_contiguous(dev, page, size >> PAGE_SHIFT);Additionally, as CMA can be used only in the context which permits sleeping, most of callers do a
gfpflags_allow_blocking()check and a corresponding fallback allocation of normal pages upon any false result:
/* A piece of example */
if (gfpflags_allow_blocking(flag))page = dma_alloc_from_contiguous(); if (!page) page = alloc_pages(); [...] if (!dma_release_from_contiguous(dev, page, count)) __free_pages(page, get_order(size));So this patch simplifies those function calls by abstracting these operations into the two new functions: dma_{alloc,free}_contiguous.
As some callers of dma_{alloc,release}_from_contiguous() might be complicated, this patch just implements these two new functions to
kernel/dma/direct.conly as an initial step.Suggested-by Christoph Hellwig <[email protected]>Signed-off-by Nicolin Chen <[email protected]>---
Changelogv2->v3:
* Added missing "static inline" in header file to fix build error.v1->v2:
* Added new functions beside the old ones so we can replace callers one by one later.
* Applied new functions todma/direct.conly, because it's the best example caller to apply and should be safe with the new functions.include/linux/dma-contiguous.h | 11 ++++++++ kernel/dma/contiguous.c | 48 ++++++++++++++++++++++++++++++++++ kernel/dma/direct.c | 24 +++-------------- 3 files changed, 63 insertions(+), 20 deletions(-)include/linux/dma-contiguous.h +11
-0@@ -115,6 +115,8 @@ struct page *dma_alloc_from_contiguous(struct device *dev, size_t count, unsigned int order, bool no_warn);bool dma_release_from_contiguous (struct device *dev,struct page *pages,int count); +struct page *dma_alloc_contiguous (struct device *dev,size_t size,gfp_t gfp); +void dma_free_contiguous (struct device *dev,struct page *page,size_t size);#else @@ -157,6 +159,15 @@ bool dma_release_from_contiguous(struct device *dev, struct page *pages,return false; } +static inline +struct page *dma_alloc_contiguous (struct device *dev,size_t size,gfp_t gfp) +{ +return NULL ; +} + +static inline +void dma_free_contiguous (struct device *dev,struct page *page,size_t size) { } +#endif #endif kernel/dma/contiguous.c +48
-0@@ -214,6 +214,54 @@ bool dma_release_from_contiguous(struct device *dev, struct page *pages,return cma_release (dev_get_cma_area (dev), pages, count); } +/** +* dma_alloc_contiguous() - allocate contiguous pages +* @dev: Pointer to device for which the allocation is performed. +* @size: Requested allocation size. +* @gfp: Allocation flags. +* +* This function allocates contiguous memory buffer for specified device. It +* first tries to use device specific contiguous memory area if available or +* the default global one, then tries a fallback allocation of normal pages. +*/ +struct page *dma_alloc_contiguous (struct device *dev,size_t size,gfp_t gfp) +{ +int node = dev ?dev_to_node (dev) :NUMA_NO_NODE ; +size_t count =PAGE_ALIGN (size) >>PAGE_SHIFT ; +size_t align =get_order (PAGE_ALIGN (size)); +struct cma *cma =dev_get_cma_area (dev); +struct page *page =NULL ; + +/* CMA can be used only in the context which permits sleeping */ +if (cma &&gfpflags_allow_blocking (gfp)) { + align =min_t (size_t, align,CONFIG_CMA_ALIGNMENT ); + page =cma_alloc (cma, count, align, gfp & __GFP_NOWARN); + } + +/* Fallback allocation of normal pages */ +if (!page) + page =alloc_pages_node (node, gfp, align); + +return page; +} + +/** +* dma_free_contiguous() - release allocated pages +* @dev: Pointer to device for which the pages were allocated. +* @page: Pointer to the allocated pages. +* @size: Size of allocated pages. +* +* This function releases memory allocated by dma_alloc_contiguous(). As the +* cma_release returns false when provided pages do not belong to contiguous +* area and true otherwise, this function then does a fallback __free_pages() +* upon a false-return. +*/ +void dma_free_contiguous (struct device *dev,struct page *page,size_t size) +{ +if (!cma_release (dev_get_cma_area (dev), page, size >>PAGE_SHIFT )) +__free_pages (page,get_order (size)); +} +/* * Support for reserved memory regions defined in device tree */ kernel/dma/direct.c +4
-21@@ -96,8 +96,6 @@ static bool dma_coherent_ok(struct device *dev, phys_addr_t phys, size_t size)struct page *__dma_direct_alloc_pages (struct device *dev,size_t size,dma_addr_t *dma_handle,gfp_t gfp,unsigned long attrs) { -unsigned int count =PAGE_ALIGN (size) >>PAGE_SHIFT ; -int page_order =get_order (size);struct page *page =NULL ;u64 phys_mask;@@ -109,20 +107,9 @@ struct page *__dma_direct_alloc_pages(struct device *dev, size_t size, gfp |=__dma_direct_optimal_gfp_mask (dev, dev->coherent_dma_mask, &phys_mask); again: -/* CMA can be used only in the context which permits sleeping */ -if (gfpflags_allow_blocking (gfp)) { - page =dma_alloc_from_contiguous (dev, count, page_order, - gfp & __GFP_NOWARN); -if (page && !dma_coherent_ok (dev,page_to_phys (page), size)) { -dma_release_from_contiguous (dev, page, count); - page =NULL ; - } - } -if (!page) - page =alloc_pages_node (dev_to_node (dev), gfp, page_order); - + page =dma_alloc_contiguous (dev, size, gfp);if (page && !dma_coherent_ok (dev,page_to_phys (page), size)) { -__free_pages (page, page_order); +dma_free_contiguous (dev, page, size); page =NULL ; if (IS_ENABLED(CONFIG_ZONE_DMA32) &&@@ -154,7 +141,7 @@ void *dma_direct_alloc_pages(struct device *dev, size_t size,if (PageHighMem (page)) {/* * Depending on the cma= arguments and per-arch setup -* dma_alloc_from_contiguous could return highmem pages. +* dma_alloc_contiguous could return highmem pages. * Without remapping there is no way to return them here, * so log an error and fail. */ @@ -176,10 +163,7 @@ void *dma_direct_alloc_pages(struct device *dev, size_t size,void __dma_direct_free_pages (struct device *dev,size_t size,struct page *page) { -unsigned int count =PAGE_ALIGN (size) >>PAGE_SHIFT ; - -if (!dma_release_from_contiguous (dev, page, count)) -__free_pages (page,get_order (size)); +dma_free_contiguous (dev, page, size); } void dma_direct_free_pages(struct device *dev, size_t size, void *cpu_addr, --2.17.1
Re: [PATCH v3 1/2] dma-contiguous: Abstract dma_{alloc,free}_contiguous()
Quoted 3 lines · Ezequiel Garcia
I can't find a way to forward-redirect from Gmail, so I'm Ccing Dafna who found a regression caused by this commit. Dafna, can you give all the details, including the log and how you are reproducing it?
I saw the conversation there. Sorry for not replying yet. May we discuss there since there are full logs available?
Thanks
Nicolin
Quoted 197 lines · Nicolin Chen
Quoted 194 lines
Both
dma_alloc_from_contiguous()anddma_release_from_contiguous()are very simply implemented, but requiring callers to pass certain parameters like count and align, and taking a boolean parameter to check__GFP_NOWARNin the allocation flags. So every function call duplicates similar work:
/* A piece of example */
unsigned long order =get_order(size);
size_tcount = size >>PAGE_SHIFT;
page =dma_alloc_from_contiguous(dev, count, order, gfp & __GFP_NOWARN); [...]
dma_release_from_contiguous(dev, page, size >> PAGE_SHIFT);Additionally, as CMA can be used only in the context which permits sleeping, most of callers do a
gfpflags_allow_blocking()check and a corresponding fallback allocation of normal pages upon any false result:
/* A piece of example */
if (gfpflags_allow_blocking(flag))page = dma_alloc_from_contiguous(); if (!page) page = alloc_pages(); [...] if (!dma_release_from_contiguous(dev, page, count)) __free_pages(page, get_order(size));So this patch simplifies those function calls by abstracting these operations into the two new functions: dma_{alloc,free}_contiguous.
As some callers of dma_{alloc,release}_from_contiguous() might be complicated, this patch just implements these two new functions to
kernel/dma/direct.conly as an initial step.Suggested-by Christoph Hellwig <[email protected]>Signed-off-by Nicolin Chen <[email protected]>---
Changelogv2->v3:
* Added missing "static inline" in header file to fix build error.v1->v2:
* Added new functions beside the old ones so we can replace callers one by one later.
* Applied new functions todma/direct.conly, because it's the best example caller to apply and should be safe with the new functions.include/linux/dma-contiguous.h | 11 ++++++++ kernel/dma/contiguous.c | 48 ++++++++++++++++++++++++++++++++++ kernel/dma/direct.c | 24 +++-------------- 3 files changed, 63 insertions(+), 20 deletions(-)include/linux/dma-contiguous.h +11
-0@@ -115,6 +115,8 @@ struct page *dma_alloc_from_contiguous(struct device *dev, size_t count, unsigned int order, bool no_warn);bool dma_release_from_contiguous (struct device *dev,struct page *pages,int count); +struct page *dma_alloc_contiguous (struct device *dev,size_t size,gfp_t gfp); +void dma_free_contiguous (struct device *dev,struct page *page,size_t size);#else @@ -157,6 +159,15 @@ bool dma_release_from_contiguous(struct device *dev, struct page *pages,return false; } +static inline +struct page *dma_alloc_contiguous (struct device *dev,size_t size,gfp_t gfp) +{ +return NULL ; +} + +static inline +void dma_free_contiguous (struct device *dev,struct page *page,size_t size) { } +#endif #endif kernel/dma/contiguous.c +48
-0@@ -214,6 +214,54 @@ bool dma_release_from_contiguous(struct device *dev, struct page *pages,return cma_release (dev_get_cma_area (dev), pages, count); } +/** +* dma_alloc_contiguous() - allocate contiguous pages +* @dev: Pointer to device for which the allocation is performed. +* @size: Requested allocation size. +* @gfp: Allocation flags. +* +* This function allocates contiguous memory buffer for specified device. It +* first tries to use device specific contiguous memory area if available or +* the default global one, then tries a fallback allocation of normal pages. +*/ +struct page *dma_alloc_contiguous (struct device *dev,size_t size,gfp_t gfp) +{ +int node = dev ?dev_to_node (dev) :NUMA_NO_NODE ; +size_t count =PAGE_ALIGN (size) >>PAGE_SHIFT ; +size_t align =get_order (PAGE_ALIGN (size)); +struct cma *cma =dev_get_cma_area (dev); +struct page *page =NULL ; + +/* CMA can be used only in the context which permits sleeping */ +if (cma &&gfpflags_allow_blocking (gfp)) { + align =min_t (size_t, align,CONFIG_CMA_ALIGNMENT ); + page =cma_alloc (cma, count, align, gfp & __GFP_NOWARN); + } + +/* Fallback allocation of normal pages */ +if (!page) + page =alloc_pages_node (node, gfp, align); + +return page; +} + +/** +* dma_free_contiguous() - release allocated pages +* @dev: Pointer to device for which the pages were allocated. +* @page: Pointer to the allocated pages. +* @size: Size of allocated pages. +* +* This function releases memory allocated by dma_alloc_contiguous(). As the +* cma_release returns false when provided pages do not belong to contiguous +* area and true otherwise, this function then does a fallback __free_pages() +* upon a false-return. +*/ +void dma_free_contiguous (struct device *dev,struct page *page,size_t size) +{ +if (!cma_release (dev_get_cma_area (dev), page, size >>PAGE_SHIFT )) +__free_pages (page,get_order (size)); +} +/* * Support for reserved memory regions defined in device tree */ kernel/dma/direct.c +4
-20@@ -96,8 +96,6 @@ static bool dma_coherent_ok(struct device *dev, phys_addr_t phys, size_t size)struct page *__dma_direct_alloc_pages (struct device *dev,size_t size,dma_addr_t *dma_handle,gfp_t gfp,unsigned long attrs) { -unsigned int count =PAGE_ALIGN (size) >>PAGE_SHIFT ; -int page_order =get_order (size);struct page *page =NULL ;u64 phys_mask;@@ -109,20 +107,9 @@ struct page *__dma_direct_alloc_pages(struct device *dev, size_t size, gfp |=__dma_direct_optimal_gfp_mask (dev, dev->coherent_dma_mask, &phys_mask); again: -/* CMA can be used only in the context which permits sleeping */ -if (gfpflags_allow_blocking (gfp)) { - page =dma_alloc_from_contiguous (dev, count, page_order, - gfp & __GFP_NOWARN); -if (page && !dma_coherent_ok (dev,page_to_phys (page), size)) { -dma_release_from_contiguous (dev, page, count); - page =NULL ; - } - } -if (!page) - page =alloc_pages_node (dev_to_node (dev), gfp, page_order); - + page =dma_alloc_contiguous (dev, size, gfp);if (page && !dma_coherent_ok (dev,page_to_phys (page), size)) { -__free_pages (page, page_order); +dma_free_contiguous (dev, page, size); page =NULL ; if (IS_ENABLED(CONFIG_ZONE_DMA32) &&@@ -154,7 +141,7 @@ void *dma_direct_alloc_pages(struct device *dev, size_t size,if (PageHighMem (page)) {/* * Depending on the cma= arguments and per-arch setup -* dma_alloc_from_contiguous could return highmem pages. +* dma_alloc_contiguous could return highmem pages. * Without remapping there is no way to return them here, * so log an error and fail. */ @@ -176,10 +163,7 @@ void *dma_direct_alloc_pages(struct device *dev, size_t size,void __dma_direct_free_pages (struct device *dev,size_t size,struct page *page) { -unsigned int count =PAGE_ALIGN (size) >>PAGE_SHIFT ; - -if (!dma_release_from_contiguous (dev, page, count)) -__free_pages (page,get_order (size)); +dma_free_contiguous (dev, page, size); }
Quoted 4 lines
Quoted 4 lines
kernel/dma/direct.c +0
-1void dma_direct_free_pages(struct device *dev, size_t size, void *cpu_addr, --2.17.1
_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
Re: [PATCH v3 1/2] dma-contiguous: Abstract dma_{alloc,free}_contiguous()
Quoted 11 lines · Nicolin Chen
Quoted 3 lines · Ezequiel Garcia
I can't find a way to forward-redirect from Gmail, so I'm Ccing Dafna who found a regression caused by this commit. Dafna, can you give all the details, including the log and how you are reproducing it?
I saw the conversation there. Sorry for not replying yet. May we discuss there since there are full logs available?
Thanks
Nicolin
Hi,
I compiled vivid as built-in into the kernel (not as a separate module) for nitrogen8m device (imx8) and set it to use contig dma for mem_ops by adding the kernel param vivid.allocators=1,1,...
I use this devicetree patch for the dtb file: https://lkml.org/lkml/2019/7/24/789. Although it should be the same on any Aarch64 platform.
Then, on the board I run the command:
v4l2-ctl -d3 -v width=2592,height=1944,pixelformat=UYVY,bytesperline=5184 --stream-mmap --stream-to video.UYVY
In every run there is a different crash. Here is one of them: https://pastebin.com/xXgbXMAN
Dafna
Quoted 201 lines
Quoted 201 lines
Quoted 201 lines
Quoted 3 lines · Nicolin Chen
Both
dma_alloc_from_contiguous()anddma_release_from_contiguous()are very simply implemented, but requiring callers to pass certain parameters like count and align, and taking a boolean parameter to check__GFP_NOWARNin the allocation flags. So every function call duplicates similar work:
/* A piece of example */
unsigned long order =get_order(size);
size_tcount = size >>PAGE_SHIFT;
page =dma_alloc_from_contiguous(dev, count, order, gfp & __GFP_NOWARN); [...]
dma_release_from_contiguous(dev, page, size >> PAGE_SHIFT);Additionally, as CMA can be used only in the context which permits sleeping, most of callers do a
gfpflags_allow_blocking()check and a corresponding fallback allocation of normal pages upon any false result:
/* A piece of example */
if (gfpflags_allow_blocking(flag))
page =dma_alloc_from_contiguous();
if (!page)
page =alloc_pages();
[...]
if (!dma_release_from_contiguous(dev, page, count))__free_pages(page, get_order(size));So this patch simplifies those function calls by abstracting these operations into the two new functions: dma_{alloc,free}_contiguous.
As some callers of dma_{alloc,release}_from_contiguous() might be complicated, this patch just implements these two new functions to
kernel/dma/direct.conly as an initial step.Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Suggested-by Christoph Hellwig <[email protected]>Signed-off-by Nicolin Chen <[email protected]>---
Changelogv2->v3:
* Added missing "static inline" in header file to fix build error.v1->v2:
* Added new functions beside the old ones so we can replace callers one by one later.
* Applied new functions todma/direct.conly, because it's the best example caller to apply and should be safe with the new functions.include/linux/dma-contiguous.h | 11 ++++++++ kernel/dma/contiguous.c | 48 ++++++++++++++++++++++++++++++++++ kernel/dma/direct.c | 24 +++--------------
3 files changed, 63 insertions(+), 20 deletions(-)include/linux/dma-contiguous.h +11
-0@@ -115,6 +115,8 @@ struct page *dma_alloc_from_contiguous(struct device *dev, size_t count, unsigned int order, bool no_warn); bool dma_release_from_contiguous(struct device *dev, struct page *pages, int count); +struct page *dma_alloc_contiguous (struct device *dev,size_t size,gfp_t gfp); +void dma_free_contiguous (struct device *dev,struct page *page,size_t size); #else@@ -157,6 +159,15 @@ bool dma_release_from_contiguous(struct device *dev, struct page *pages, return false; } +static inline +struct page *dma_alloc_contiguous (struct device *dev,size_t size,gfp_t gfp) +{ + return NULL; +} + +static inline +void dma_free_contiguous (struct device *dev,struct page *page,size_t size) { } + #endif #endifkernel/dma/contiguous.c +48
-0@@ -214,6 +214,54 @@ bool dma_release_from_contiguous(struct device *dev, struct page *pages, return cma_release(dev_get_cma_area(dev), pages, count); } +/** +* dma_alloc_contiguous() - allocate contiguous pages +* @dev: Pointer to device for which the allocation is performed. +* @size: Requested allocation size. +* @gfp: Allocation flags. +* +* This function allocates contiguous memory buffer for specified device. It +* first tries to use device specific contiguous memory area if available or +* the default global one, then tries a fallback allocation of normal pages. +*/ +struct page *dma_alloc_contiguous (struct device *dev,size_t size,gfp_t gfp) +{ + int node = dev ? dev_to_node(dev) : NUMA_NO_NODE; + size_t count = PAGE_ALIGN(size) >> PAGE_SHIFT; + size_t align = get_order(PAGE_ALIGN(size)); + struct cma *cma = dev_get_cma_area(dev); + struct page *page = NULL; + + /* CMA can be used only in the context which permits sleeping */ + if (cma && gfpflags_allow_blocking(gfp)) { + align = min_t(size_t, align, CONFIG_CMA_ALIGNMENT); + page = cma_alloc(cma, count, align, gfp & __GFP_NOWARN); + } + + /* Fallback allocation of normal pages */ + if (!page) + page = alloc_pages_node(node, gfp, align); + + return page; +} + +/** +* dma_free_contiguous() - release allocated pages +* @dev: Pointer to device for which the pages were allocated. +* @page: Pointer to the allocated pages. +* @size: Size of allocated pages. +* +* This function releases memory allocated by dma_alloc_contiguous(). As the +* cma_release returns false when provided pages do not belong to contiguous +* area and true otherwise, this function then does a fallback __free_pages() +* upon a false-return. +*/ +void dma_free_contiguous (struct device *dev,struct page *page,size_t size) +{ + if (!cma_release(dev_get_cma_area(dev), page, size >> PAGE_SHIFT)) + __free_pages(page, get_order(size)); +} + /** Support for reserved memory regions defined in device tree */ kernel/dma/direct.c +0
-2@@ -96,8 +96,6 @@ static bool dma_coherent_ok(struct device *dev, phys_addr_t phys, size_t size) struct page *__dma_direct_alloc_pages(struct device *dev, size_t size, dma_addr_t *dma_handle, gfp_t gfp, unsigned long attrs) { - unsigned int count = PAGE_ALIGN(size) >> PAGE_SHIFT; - int page_order = get_order(size); struct page *page = NULL; u64 phys_mask;@@ -109,20 +107,9 @@ struct page *__dma_direct_alloc_pages(struct device *dev, size_t size, gfp |= __dma_direct_optimal_gfp_mask(dev, dev->coherent_dma_mask, &phys_mask);again:
- /* CMA can be used only in the context which permits sleeping */ - if (gfpflags_allow_blocking(gfp)) {
- page = dma_alloc_from_contiguous(dev, count,page_order, - gfp &__GFP_NOWARN); - if (page && !dma_coherent_ok(dev,page_to_phys(page), size)) { - dma_release_from_contiguous(dev, page, count); - page = NULL;
- }
- }
- if (!page)
- page =alloc_pages_node(dev_to_node(dev), gfp, page_order); -
+ page =dma_alloc_contiguous(dev, size, gfp);
if (page && !dma_coherent_ok(dev,page_to_phys(page), size)) { - __free_pages(page,page_order);
+ dma_free_contiguous(dev, page, size);
page = NULL;if (
IS_ENABLED(CONFIG_ZONE_DMA32)&&kernel/dma/direct.c +2
-6@@ -154,7 +141,7 @@ void *dma_direct_alloc_pages(struct device *dev, size_t size, if (PageHighMem(page)) { /** Depending on the cma= arguments and per-arch setup -* dma_alloc_from_contiguous could return highmem pages. +* dma_alloc_contiguous could return highmem pages. * Without remapping there is no way to return them here, * so log an error and fail. */ @@ -176,10 +163,7 @@ void *dma_direct_alloc_pages(struct device *dev, size_t size, void __dma_direct_free_pages(struct device *dev, size_t size, struct page *page) { - unsigned int count = PAGE_ALIGN(size) >> PAGE_SHIFT; - - if (!dma_release_from_contiguous(dev, page, count)) - __free_pages(page, get_order(size)); + dma_free_contiguous(dev, page, size); } void dma_direct_free_pages(struct device *dev, size_t size, void *cpu_addr, --2.17.1
Re: [PATCH v3 1/2] dma-contiguous: Abstract dma_{alloc,free}_contiguous()
Quoted 18 lines · Dafna Hirschfeld
Quoted 11 lines · Nicolin Chen
Quoted 3 lines · Ezequiel Garcia
I can't find a way to forward-redirect from Gmail, so I'm Ccing Dafna who found a regression caused by this commit. Dafna, can you give all the details, including the log and how you are reproducing it?
I saw the conversation there. Sorry for not replying yet. May we discuss there since there are full logs available?
Thanks
NicolinHi,
I compiled vivid as built-in into the kernel (not as a separate module) for nitrogen8m device (imx8) and set it to use contig dma formem_opsby adding the kernel param vivid.allocators=1,1,...
Quoted 8 lines
I use this devicetree patch for the dtb file: https://lkml.org/lkml/2019/7/24/789. Although it should be the same on any Aarch64 platform.
Then, on the board I run the command:
v4l2-ctl -d3 -v width=2592,height=1944,pixelformat=UYVY,bytesperline=5184 --stream-mmap --stream-to video.UYVY
In every run there is a different crash. Here is one of them: https://pastebin.com/xXgbXMAN
This probably should be a fix: https://lkml.org/lkml/2019/7/25/1432
I also sent it to you. Would it be possible for you to test it?
Thanks
Nicolin
Quoted 199 lines
Dafna
Quoted 197 lines
Quoted 197 lines
Quoted 197 lines
Quoted 3 lines · Nicolin Chen
Both
dma_alloc_from_contiguous()anddma_release_from_contiguous()are very simply implemented, but requiring callers to pass certain parameters like count and align, and taking a boolean parameter to check__GFP_NOWARNin the allocation flags. So every function call duplicates similar work:
/* A piece of example */
unsigned long order =get_order(size);
size_tcount = size >>PAGE_SHIFT;
page =dma_alloc_from_contiguous(dev, count, order, gfp & __GFP_NOWARN); [...]
dma_release_from_contiguous(dev, page, size >> PAGE_SHIFT);Additionally, as CMA can be used only in the context which permits sleeping, most of callers do a
gfpflags_allow_blocking()check and a corresponding fallback allocation of normal pages upon any false result:
/* A piece of example */
if (gfpflags_allow_blocking(flag))
page =dma_alloc_from_contiguous();
if (!page)
page =alloc_pages();
[...]
if (!dma_release_from_contiguous(dev, page, count))__free_pages(page, get_order(size));So this patch simplifies those function calls by abstracting these operations into the two new functions: dma_{alloc,free}_contiguous.
As some callers of dma_{alloc,release}_from_contiguous() might be complicated, this patch just implements these two new functions to
kernel/dma/direct.conly as an initial step.Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Suggested-by Christoph Hellwig <[email protected]>Signed-off-by Nicolin Chen <[email protected]>---
Changelogv2->v3:
* Added missing "static inline" in header file to fix build error.v1->v2:
* Added new functions beside the old ones so we can replace callers one by one later.
* Applied new functions todma/direct.conly, because it's the best example caller to apply and should be safe with the new functions.include/linux/dma-contiguous.h | 11 ++++++++ kernel/dma/contiguous.c | 48 ++++++++++++++++++++++++++++++++++ kernel/dma/direct.c | 24 +++--------------
3 files changed, 63 insertions(+), 20 deletions(-)include/linux/dma-contiguous.h +11
-0@@ -115,6 +115,8 @@ struct page *dma_alloc_from_contiguous(struct device *dev, size_t count, unsigned int order, bool no_warn); bool dma_release_from_contiguous(struct device *dev, struct page *pages, int count); +struct page *dma_alloc_contiguous (struct device *dev,size_t size,gfp_t gfp); +void dma_free_contiguous (struct device *dev,struct page *page,size_t size); #else@@ -157,6 +159,15 @@ bool dma_release_from_contiguous(struct device *dev, struct page *pages, return false; } +static inline +struct page *dma_alloc_contiguous (struct device *dev,size_t size,gfp_t gfp) +{ + return NULL; +} + +static inline +void dma_free_contiguous (struct device *dev,struct page *page,size_t size) { } + #endif #endifkernel/dma/contiguous.c +48
-0@@ -214,6 +214,54 @@ bool dma_release_from_contiguous(struct device *dev, struct page *pages, return cma_release(dev_get_cma_area(dev), pages, count); } +/** +* dma_alloc_contiguous() - allocate contiguous pages +* @dev: Pointer to device for which the allocation is performed. +* @size: Requested allocation size. +* @gfp: Allocation flags. +* +* This function allocates contiguous memory buffer for specified device. It +* first tries to use device specific contiguous memory area if available or +* the default global one, then tries a fallback allocation of normal pages. +*/ +struct page *dma_alloc_contiguous (struct device *dev,size_t size,gfp_t gfp) +{ + int node = dev ? dev_to_node(dev) : NUMA_NO_NODE; + size_t count = PAGE_ALIGN(size) >> PAGE_SHIFT; + size_t align = get_order(PAGE_ALIGN(size)); + struct cma *cma = dev_get_cma_area(dev); + struct page *page = NULL; + + /* CMA can be used only in the context which permits sleeping */ + if (cma && gfpflags_allow_blocking(gfp)) { + align = min_t(size_t, align, CONFIG_CMA_ALIGNMENT); + page = cma_alloc(cma, count, align, gfp & __GFP_NOWARN); + } + + /* Fallback allocation of normal pages */ + if (!page) + page = alloc_pages_node(node, gfp, align); + + return page; +} + +/** +* dma_free_contiguous() - release allocated pages +* @dev: Pointer to device for which the pages were allocated. +* @page: Pointer to the allocated pages. +* @size: Size of allocated pages. +* +* This function releases memory allocated by dma_alloc_contiguous(). As the +* cma_release returns false when provided pages do not belong to contiguous +* area and true otherwise, this function then does a fallback __free_pages() +* upon a false-return. +*/ +void dma_free_contiguous (struct device *dev,struct page *page,size_t size) +{ + if (!cma_release(dev_get_cma_area(dev), page, size >> PAGE_SHIFT)) + __free_pages(page, get_order(size)); +} + /** Support for reserved memory regions defined in device tree */ kernel/dma/direct.c +0
-2@@ -96,8 +96,6 @@ static bool dma_coherent_ok(struct device *dev, phys_addr_t phys, size_t size) struct page *__dma_direct_alloc_pages(struct device *dev, size_t size, dma_addr_t *dma_handle, gfp_t gfp, unsigned long attrs) { - unsigned int count = PAGE_ALIGN(size) >> PAGE_SHIFT; - int page_order = get_order(size); struct page *page = NULL; u64 phys_mask;@@ -109,20 +107,9 @@ struct page *__dma_direct_alloc_pages(struct device *dev, size_t size, gfp |= __dma_direct_optimal_gfp_mask(dev, dev->coherent_dma_mask, &phys_mask);again:
- /* CMA can be used only in the context which permits sleeping */ - if (gfpflags_allow_blocking(gfp)) {
- page = dma_alloc_from_contiguous(dev, count,page_order, - gfp &__GFP_NOWARN); - if (page && !dma_coherent_ok(dev,page_to_phys(page), size)) { - dma_release_from_contiguous(dev, page, count); - page = NULL;
- }
- }
- if (!page)
- page =alloc_pages_node(dev_to_node(dev), gfp, page_order); -
+ page =dma_alloc_contiguous(dev, size, gfp);
if (page && !dma_coherent_ok(dev,page_to_phys(page), size)) { - __free_pages(page,page_order);
+ dma_free_contiguous(dev, page, size);
page = NULL;if (
IS_ENABLED(CONFIG_ZONE_DMA32)&&kernel/dma/direct.c +2
-5@@ -154,7 +141,7 @@ void *dma_direct_alloc_pages(struct device *dev, size_t size, if (PageHighMem(page)) { /** Depending on the cma= arguments and per-arch setup -* dma_alloc_from_contiguous could return highmem pages. +* dma_alloc_contiguous could return highmem pages. * Without remapping there is no way to return them here, * so log an error and fail. */ @@ -176,10 +163,7 @@ void *dma_direct_alloc_pages(struct device *dev, size_t size, void __dma_direct_free_pages(struct device *dev, size_t size, struct page *page) { - unsigned int count = PAGE_ALIGN(size) >> PAGE_SHIFT; - - if (!dma_release_from_contiguous(dev, page, count)) - __free_pages(page, get_order(size)); + dma_free_contiguous(dev, page, size); }
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
kernel/dma/direct.c +0
-1void dma_direct_free_pages(struct device *dev, size_t size, void *cpu_addr, --2.17.1
_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
PATCH v3 2/2 dma-contiguous: Use fallback alloc_pages for single pages
In reply to Nicolin ChenThe addresses within a single page are always contiguous, so it's not so necessary to always allocate one single page from CMA area. Since the CMA area has a limited predefined size of space, it may run out of space in heavy use cases, where there might be quite a lot CMA pages being allocated for single pages.
However, there is also a concern that a device might care where a page comes from -- it might expect the page from CMA area and act differently if the page doesn't.
This patch tries to use the fallback alloc_pages path, instead of one-page size allocations from the global CMA area in case that a device does not have its own CMA area. This'd save resources from the CMA global area for more CMA allocations, and also reduce CMA fragmentations resulted from trivial allocations.
--- kernel/dma/contiguous.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-)
kernel/dma/contiguous.c +10 -1
@@ -223,14 +223,23 @@ bool dma_release_from_contiguous(struct device *dev, struct page *pages,* This function allocates contiguous memory buffer for specified device. It * first tries to use device specific contiguous memory area if available or * the default global one, then tries a fallback allocation of normal pages. +* +* Note that it byapss one-page size of allocations from the global area as +* the addresses within one page are always contiguous, so there is no need +* to waste CMA pages for that kind; it also helps reduce fragmentations. */ struct page *dma_alloc_contiguous (struct device *dev,size_t size,gfp_t gfp) {int node = dev ?dev_to_node (dev) :NUMA_NO_NODE ;size_t count =PAGE_ALIGN (size) >>PAGE_SHIFT ;size_t align =get_order (PAGE_ALIGN (size)); -struct cma *cma =dev_get_cma_area (dev);struct page *page =NULL ; +struct cma *cma =NULL ; + +if (dev && dev->cma_area) + cma = dev->cma_area; +else if (count >1 ) + cma = dma_contiguous_default_area;/* CMA can be used only in the context which permits sleeping */ if (cma &&gfpflags_allow_blocking (gfp)) {
Signature · 1 line
2.17.1
Re: [PATCH v3 2/2] dma-contiguous: Use fallback alloc_pages for single pages
Quoted 9 lines · Nicolin Chen
The addresses within a single page are always contiguous, so it's not so necessary to always allocate one single page from CMA area. Since the CMA area has a limited predefined size of space, it may run out of space in heavy use cases, where there might be quite a lot CMA pages being allocated for single pages.
However, there is also a concern that a device might care where a page comes from -- it might expect the page from CMA area and act differently if the page doesn't.
How does a device know, after this call, if a CMA area was used? From the patches I figured a device should not care.
Quoted 28 lines
This patch tries to use the fallback
alloc_pagespath, instead of one-page size allocations from the global CMA area in case that a device does not have its own CMA area. This'd save resources from the CMA global area for more CMA allocations, and also reduce CMA fragmentations resulted from trivial allocations.Signed-off-by Nicolin Chen <[email protected]>--- kernel/dma/contiguous.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-)kernel/dma/contiguous.c +4
-0@@ -223,14 +223,23 @@ bool dma_release_from_contiguous(struct device *dev, struct page *pages,* This function allocates contiguous memory buffer for specified device. It * first tries to use device specific contiguous memory area if available or * the default global one, then tries a fallback allocation of normal pages. +* +* Note that it byapss one-page size of allocations from the global area as +* the addresses within one page are always contiguous, so there is no need +* to waste CMA pages for that kind; it also helps reduce fragmentations. */ struct page *dma_alloc_contiguous (struct device *dev,size_t size,gfp_t gfp) {int node = dev ?dev_to_node (dev) :NUMA_NO_NODE ;
kernel/dma/contiguous.c bool dma_release_from_contiguous(struct device *dev, struct page *pages, · line 242
Quoted 10 lines
kernel/dma/contiguous.c +6
-1size_t count =PAGE_ALIGN (size) >>PAGE_SHIFT ;size_t align =get_order (PAGE_ALIGN (size)); -struct cma *cma =dev_get_cma_area (dev);struct page *page =NULL ; +struct cma *cma =NULL ; + +if (dev && dev->cma_area) + cma = dev->cma_area; +else if (count >1 ) + cma = dma_contiguous_default_area;
Doesn't dev_get_dma_area() already do this?
Ira
Quoted 6 lines
kernel/dma/contiguous.c +0
-0/* CMA can be used only in the context which permits sleeping */ if (cma &&gfpflags_allow_blocking (gfp)) {Signature · 1 line
2.17.1
Re: [PATCH v3 2/2] dma-contiguous: Use fallback alloc_pages for single pages
Hi Ira,
Quoted 7 lines · Ira Weiny
Quoted 6 lines · Nicolin Chen
The addresses within a single page are always contiguous, so it's not so necessary to always allocate one single page from CMA area. Since the CMA area has a limited predefined size of space, it may run out of space in heavy use cases, where there might be quite a lot CMA pages being allocated for single pages.
Quoted 6 lines
Quoted 3 lines
However, there is also a concern that a device might care where a page comes from -- it might expect the page from CMA area and act differently if the page doesn't.
How does a device know, after this call, if a CMA area was used? From the patches I figured a device should not care.
A device doesn't know. But that doesn't mean a device won't care at all. There was a concern from Robin and Christoph, as a corner case that device might act differently if the memory isn't in its own CMA region. That's why we let it still use its device specific CMA area.
Quoted 6 lines
kernel/dma/contiguous.c bool dma_release_from_contiguous(struct device *dev, struct page *pages, · line 242
Quoted 4 lines
kernel/dma/contiguous.c +4
-0+if (dev && dev->cma_area) + cma = dev->cma_area; +else if (count >1 ) + cma = dma_contiguous_default_area;Doesn't
dev_get_dma_area()already do this?
Partially yes. But unwrapping it makes the program flow clear in my opinion. Actually I should have mentioned that this patch was suggested by Christoph also.
Otherwise, it would need an override like:
cma = dev_get_dma_area(); if (count > 1 && cma == dma_contiguous_default_area) cma = NULL;
Which doesn't look that bad though..
Thanks
Nicolin
Quoted 13 lines · Nicolin Chen
[ Per discussion at v1, we decide to add two new functions and start
replacing callers one by one. For this series, it only touches the dma-direct part. And instead of merging two PATCHes, I still keep them separate so that we may easily revert PATCH-2 if anything bad happens as last time -- PATCH-1 is supposed to be a safe cleanup. ]This series of patches try to optimize dma_*_from_contiguous calls:
PATCH-1 abstracts two new functions and applies todma-direct.cfile. PATCH-2 saves single pages and reduce fragmentations from CMA area.Please check their commit messages for detail changelog.
Quoted 8 lines
Nicolin Chen (2):
dma-contiguous: Abstract dma_{alloc,free}_contiguous()
dma-contiguous: Use fallbackalloc_pagesfor single pagesinclude/linux/dma-contiguous.h | 11 +++++++ kernel/dma/contiguous.c | 57 ++++++++++++++++++++++++++++++++++ kernel/dma/direct.c | 24 +++----------- 3 files changed, 72 insertions(+), 20 deletions(-)
Thanks Nicolin. Tested on a HiSilicon D06 system.
_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
Christoph Hellwig
AppliedThanks,
applied to dma-mapping for-next.
Can you also send a conversion of drivers/iommu/dma-iommu.c to your new helpers against this tree?
http://git.infradead.org/users/hch/dma-mapping.git/shortlog/refs/heads/for-next
Quoted 8 lines · Christoph Hellwig
Thanks,
applied to dma-mapping for-next.
Can you also send a conversion of
drivers/iommu/dma-iommu.cto your new helpers against this tree?http://git.infradead.org/users/hch/dma-mapping.git/shortlog/refs/heads/for-next
I can. There is a reported regression with !CONFIG_DMA_CMA now so I will do that after a fix is merged and the whole thing is stable.
Thank you