Fix userfaultfd_api to return EINVAL as expected
Subsystems (5)
PATCH 1/2 Fix userfaultfd_api to return EINVAL as expected
Currently if we request a feature that is not set in the Kernel config we fail silently and return the available features. However, the documentation indicates we should return an EINVAL.
We need to fix this issue since we can end up with a Kernel warning should a program request the feature UFFD_FEATURE_WP_UNPOPULATED on a kernel with the config not set with this feature.
--- fs/userfaultfd.c | 5 +++++ 1 file changed, 5 insertions(+)
fs/userfaultfd.c +5 -0
@@ -2073,6 +2073,11 @@ static int userfaultfd_api(struct userfaultfd_ctx *ctx, uffdio_api.features &= ~UFFD_FEATURE_WP_UNPOPULATED ; uffdio_api.features &= ~UFFD_FEATURE_WP_ASYNC ;#endif + + ret = -EINVAL ; +if (features & ~uffdio_api.features) + goto err_out; + uffdio_api.ioctls =UFFD_API_IOCTLS ; ret = -EFAULT ;if (copy_to_user (buf, &uffdio_api,sizeof (uffdio_api)))
Signature · 1 line
2.44.0
PATCH 2/2 Update uffd-stress to handle EINVAL for unset config features
Now that we have updated userfaultfd_api to correctly return EIVAL when a feature is requested but not available, let's fix the uffd-stress test to only set the UFFD_FEATURE_WP_UNPOPULATED feature when the config is set. In addition, still run the test if the CONFIG_PTE_MARKER_UFFD_WP is not set, just dont use the corresponding UFFD_FEATURE_WP_UNPOPULATED feature.
--- tools/testing/selftests/mm/uffd-stress.c | 14 +++++++++----- 1 file changed, 9 insertions(+), 5 deletions(-)
tools/testing/selftests/mm/uffd-stress.c +9 -5
@@ -38,6 +38,8 @@#ifdef __NR_userfaultfd +uint64_t features; +#define BOUNCE_RANDOM (1<<0)#define BOUNCE_RACINGFAULTS (1<<1)#define BOUNCE_VERIFY (1<<2)
@@ -247,10 +249,14 @@ static int userfaultfd_stress(void)unsigned long nr;struct uffd_args args[nr_cpus];uint64_t mem_size = nr_pages * page_size; +int flags =0 ;memset (args,0 ,sizeof (struct uffd_args ) * nr_cpus); -if (uffd_test_ctx_init (UFFD_FEATURE_WP_UNPOPULATED ,NULL )) +if (features &UFFD_FEATURE_WP_UNPOPULATED && test_type ==TEST_ANON ) + flags =UFFD_FEATURE_WP_UNPOPULATED ; + +if (uffd_test_ctx_init (flags,NULL ))err ("context init failed" );if (posix_memalign (&area, page_size, page_size))
@@ -385,8 +391,6 @@ static void set_test_type(const char *type)static void parse_test_type_arg (const char *raw_type) { -uint64_t features =UFFD_API_FEATURES ; -set_test_type (raw_type);if (!test_type)
@@ -409,8 +413,8 @@ static void parse_test_type_arg(const char *raw_type)* feature. */ -if (userfaultfd_open (&features)) -err ("Userfaultfd open failed" ); +if (uffd_get_features (&features)) +err ("failed to get available features" ); test_uffdio_wp = test_uffdio_wp && (features &UFFD_FEATURE_PAGEFAULT_FLAG_WP );
Signature · 1 line
2.44.0
Quoted 3 lines · Audra Mitchell
Currently if we request a feature that is not set in the Kernel config we fail silently and return the available features. However, the documentation indicates we should return an EINVAL.
I assume you are referencing
"EINVAL The API version requested in the api field is not supported by this kernel, or the features field passed to the kernel includes feature bits that are not supported by the current kernel version."
and
"To enable userfaultfd features the application should set a bit corresponding to each feature it wants to enable in the features field. If the kernel supports all the requested features it will enable them. Otherwise it will zero out the returned uffdio_api structure and return EINVAL. "
in which case I agree.
Quoted 4 lines
We need to fix this issue since we can end up with a Kernel warning should a program request the feature
UFFD_FEATURE_WP_UNPOPULATEDon a kernel with the config not set with this feature.
Can you mention which exact one? Is it a WARN* or a pr_warn() ?
Likely we want "Fixes:" here.
Quoted 12 lines
Signed-off-by Audra Mitchell <[email protected]>--- fs/userfaultfd.c | 5 +++++ 1 file changed, 5 insertions(+)fs/userfaultfd.c +0
-0@@ -2073,6 +2073,11 @@ static int userfaultfd_api(struct userfaultfd_ctx *ctx, uffdio_api.features &= ~UFFD_FEATURE_WP_UNPOPULATED ;
fs/userfaultfd.c static int userfaultfd_api(struct userfaultfd_ctx *ctx, · line 2083
Quoted 10 lines
fs/userfaultfd.c +5
-0uffdio_api.features &= ~UFFD_FEATURE_WP_ASYNC ;#endif + + ret = -EINVAL ; +if (features & ~uffdio_api.features) + goto err_out; + uffdio_api.ioctls =UFFD_API_IOCTLS ; ret = -EFAULT ;if (copy_to_user (buf, &uffdio_api,sizeof (uffdio_api)))
CCing Peter.
Signature · 3 lines
Cheers, David / dhildenb
Quoted 13 lines · David Hildenbrand
Quoted 3 lines · Audra Mitchell
Currently if we request a feature that is not set in the Kernel config we fail silently and return the available features. However, the documentation indicates we should return an EINVAL.
I assume you are referencing
"EINVAL The API version requested in the api field is not supported by this kernel, or the features field passed to the kernel includes feature bits that are not supported by the current kernel version."
and
Quoted 8 lines
"To enable userfaultfd features the application should set a bit corresponding to each feature it wants to enable in the features field. If the kernel supports all the requested features it will enable them. Otherwise it will zero out the returned
uffdio_apistructure and return EINVAL.
"in which case I agree.
Yep! I'm referencing the man page.
Quoted 7 lines
Quoted 5 lines
We need to fix this issue since we can end up with a Kernel warning should a program request the feature
UFFD_FEATURE_WP_UNPOPULATEDon a kernel with the config not set with this feature.Can you mention which exact one? Is it a WARN* or a
pr_warn()?
Here is the kernel warning I get:
[ 200.803094] unrecognized swap entry 0x7c00000000000001 [ 200.808270] ------------[ cut here ]------------ [ 200.812896] WARNING: CPU: 91 PID: 13634 at mm/memory.c:1660 zap_pte_range+0x43d/0x660 [ 200.820738] Modules linked in: qrtr bridge stp llc rfkill sunrpc amd_atl intel_rapl_msr intel_rapl_common amd64_edac edac_mce_amd kvm_amd kvm ipmi_ssif acpi_ipmi i2c_piix4 ipmi_si wmi_bmof dcdbas dell_smbios dell_wmi_descriptor ptdma ipmi_devintf rapl ipmi_msghandler acpi_power_meter pcspkr k10temp xfs libcrc32c sd_mod t10_pi mgag200 sg drm_kms_helper crct10dif_pclmul i2c_algo_bit ahci crc32_pclmul drm_shmem_helper libahci crc32c_intel drm i40e libata ghash_clmulni_intel tg3 ccp megaraid_sas sp5100_tco wmi dm_mirror dm_region_hash dm_log dm_mod fuse [ 200.869387] CPU: 91 PID: 13634 Comm: userfaultfd Kdump: loaded Not tainted 6.9.0-rc5+ #8 [ 200.877477] Hardware name: Dell Inc. PowerEdge R6525/0N7YGH, BIOS 2.7.3 03/30/2022 [ 200.885052] RIP: 0010:zap_pte_range+0x43d/0x660 [ 200.889595] Code: 83 fa 02 0f 86 44 01 00 00 83 f9 17 0f 84 e1 00 00 00 83 f9 1f 0f 84 d0 00 00 00 48 89 c6 48 c7 c7 00 e4 dd bb e8 73 a2 de ff <0f> 0b e9 44 fd ff ff 45 0f b6 44 24 20 41 f6 c0 f4 75 27 4c 89 ee [ 200.908348] RSP: 0018:ffffa18d2e6c37c8 EFLAGS: 00010246 [ 200.913584] RAX: 000000000000002a RBX: 00007f26d3600000 RCX: 0000000000000000 [ 200.920730] RDX: 0000000000000000 RSI: ffff93503f9a0bc0 RDI: ffff93503f9a0bc0 [ 200.927867] RBP: 00007f26d35cc000 R08: 0000000000000000 R09: ffffa18d2e6c3688 [ 200.935009] R10: ffffa18d2e6c3680 R11: ffffffffbc9de448 R12: ffffa18d2e6c39e8 [ 200.942149] R13: ffff92d1ebc15b50 R14: ffff93114e0cde60 R15: ffffa18d2e6c3928 [ 200.949291] FS: 0000000000000000(0000) GS:ffff93503f980000(0000) knlGS:0000000000000000 [ 200.957384] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 200.963140] CR2: 00007f26b1600658 CR3: 00000040905ba000 CR4: 0000000000350ef0 [ 200.970283] Call Trace: [ 200.972745] <TASK> [ 200.974862] ? __warn+0x7f/0x130 [ 200.978108] ? zap_pte_range+0x43d/0x660 [ 200.982044] ? report_bug+0x18a/0x1a0 [ 200.985720] ? handle_bug+0x3c/0x70 [ 200.989219] ? exc_invalid_op+0x14/0x70 [ 200.993068] ? asm_exc_invalid_op+0x16/0x20 [ 200.997265] ? zap_pte_range+0x43d/0x660 [ 201.001199] ? zap_pte_range+0x43d/0x660 [ 201.005134] zap_pmd_range.isra.0+0xf9/0x230 [ 201.009416] unmap_page_range+0x2d4/0x4a0 [ 201.013436] unmap_vmas+0xa8/0x180 [ 201.016854] exit_mmap+0xea/0x3b0 [ 201.020191] __mmput+0x43/0x120 [ 201.023342] exit_mm+0xb1/0x110 [ 201.026496] do_exit+0x270/0x4f0 [ 201.029739] do_group_exit+0x2c/0x80 [ 201.033326] get_signal+0x886/0x8b0 [ 201.036828] ? srso_return_thunk+0x5/0x5f [ 201.040848] arch_do_signal_or_restart+0x25/0x100 [ 201.045563] ? srso_return_thunk+0x5/0x5f [ 201.049583] ? vma_set_page_prot+0x5e/0xc0 [ 201.053692] ? srso_return_thunk+0x5/0x5f [ 201.057713] ? syscall_exit_work+0xff/0x130 [ 201.061908] syscall_exit_to_user_mode+0x1b3/0x200 [ 201.066712] do_syscall_64+0x87/0x160 [ 201.070387] ? srso_return_thunk+0x5/0x5f [ 201.074405] ? do_mmap+0x416/0x5f0 [ 201.077821] ? srso_return_thunk+0x5/0x5f [ 201.081840] ? rseq_get_rseq_cs+0x1d/0x240 [ 201.085950] ? srso_return_thunk+0x5/0x5f [ 201.089970] ? rseq_ip_fixup+0x6d/0x1d0 [ 201.093823] ? vm_mmap_pgoff+0x117/0x1a0 [ 201.097755] ? srso_return_thunk+0x5/0x5f [ 201.101776] ? srso_return_thunk+0x5/0x5f [ 201.105795] ? syscall_exit_to_user_mode+0x78/0x200 [ 201.110685] ? srso_return_thunk+0x5/0x5f [ 201.114706] ? do_syscall_64+0x87/0x160 [ 201.118557] ? srso_return_thunk+0x5/0x5f [ 201.122575] ? __count_memcg_events+0x49/0xb0 [ 201.126944] ? srso_return_thunk+0x5/0x5f [ 201.130967] ? srso_return_thunk+0x5/0x5f [ 201.134986] ? syscall_exit_work+0xff/0x130 [ 201.139184] ? srso_return_thunk+0x5/0x5f [ 201.143205] ? syscall_exit_to_user_mode+0x78/0x200 [ 201.148093] ? srso_return_thunk+0x5/0x5f [ 201.152114] ? do_syscall_64+0x87/0x160 [ 201.155960] ? srso_return_thunk+0x5/0x5f [ 201.159984] ? sched_clock_cpu+0xb/0x190 [ 201.163916] ? srso_return_thunk+0x5/0x5f [ 201.167939] ? irqtime_account_irq+0x40/0xc0 [ 201.172220] ? srso_return_thunk+0x5/0x5f [ 201.176243] ? srso_return_thunk+0x5/0x5f [ 201.180263] entry_SYSCALL_64_after_hwframe+0x76/0x7e [ 201.185326] RIP: 0033:0x7f26dfd0735b [ 201.188939] Code: Unable to access opcode bytes at 0x7f26dfd07331. [ 201.195128] RSP: 002b:00007fffce176868 EFLAGS: 00000206 ORIG_RAX: 000000000000000a [ 201.202700] RAX: fffffffffffffffc RBX: 00007f26dfe60000 RCX: 00007f26dfd0735b [ 201.209841] RDX: 0000000000000003 RSI: 0000000001000000 RDI: 00007f26af401000 [ 201.216983] RBP: 00007f26b0400640 R08: 00000000ffffffff R09: 0000000000000000 [ 201.224127] R10: ffffffffffffffc0 R11: 0000000000000206 R12: 0000000000000000 [ 201.231267] R13: 000000000040d320 R14: 0000000000000000 R15: 0000000000000000 [ 201.238413] </TASK> [ 201.240610] ---[ end trace 0000000000000000 ]--- [ 201.245250] unrecognized swap entry 0x7c00000000000001
Quoted 2 lines
Likely we want "Fixes:" here.
This could be seen as a continuation of the problem2ff559f31a5d Revert "userfaultfd: don't fail on unrecognized features" was trying to solve. However, this patch only checks to make sure we didnt ask for a feature outside the possible range of features. We are still missing a check to confirm the requested features are also configured on. So I guess the "Fixes" tag would be for this patch?914eedcb9ba0 userfaultfd: don't fail on unrecognized features
Happy to get your input here!
Thanks in advance!
Quoted 31 lines
Quoted 13 lines
Signed-off-by Audra Mitchell <[email protected]>--- fs/userfaultfd.c | 5 +++++ 1 file changed, 5 insertions(+)fs/userfaultfd.c +0
-0@@ -2073,6 +2073,11 @@ static int userfaultfd_api(struct userfaultfd_ctx *ctx, uffdio_api.features &= ~UFFD_FEATURE_WP_UNPOPULATED ;fs/userfaultfd.c static int userfaultfd_api(struct userfaultfd_ctx *ctx, · line 2083
Quoted 10 lines
fs/userfaultfd.c +5
-0uffdio_api.features &= ~UFFD_FEATURE_WP_ASYNC ;#endif + + ret = -EINVAL ; +if (features & ~uffdio_api.features) + goto err_out; + uffdio_api.ioctls =UFFD_API_IOCTLS ; ret = -EFAULT ;if (copy_to_user (buf, &uffdio_api,sizeof (uffdio_api)))CCing Peter.
Signature · 3 lines
Cheers, David / dhildenb
Peter Xu
reviewerQuoted 124 lines · Audra Mitchell
Quoted 13 lines · David Hildenbrand
Quoted 3 lines · Audra Mitchell
Currently if we request a feature that is not set in the Kernel config we fail silently and return the available features. However, the documentation indicates we should return an EINVAL.
I assume you are referencing
"EINVAL The API version requested in the api field is not supported by this kernel, or the features field passed to the kernel includes feature bits that are not supported by the current kernel version."
and
Quoted 8 lines
"To enable userfaultfd features the application should set a bit corresponding to each feature it wants to enable in the features field. If the kernel supports all the requested features it will enable them. Otherwise it will zero out the returned
uffdio_apistructure and return EINVAL.
"in which case I agree.
Yep! I'm referencing the man page.
Quoted 7 lines
Quoted 5 lines
We need to fix this issue since we can end up with a Kernel warning should a program request the feature
UFFD_FEATURE_WP_UNPOPULATEDon a kernel with the config not set with this feature.Can you mention which exact one? Is it a WARN* or a
pr_warn()?Here is the kernel warning I get:
[ 200.803094] unrecognized swap entry 0x7c00000000000001 [ 200.808270] ------------[ cut here ]------------ [ 200.812896] WARNING: CPU: 91 PID: 13634 at mm/memory.c:1660 zap_pte_range+0x43d/0x660 [ 200.820738] Modules linked in: qrtr bridge stp llc rfkill sunrpc amd_atl intel_rapl_msr intel_rapl_common amd64_edac edac_mce_amd kvm_amd kvm ipmi_ssif acpi_ipmi i2c_piix4 ipmi_si wmi_bmof dcdbas dell_smbios dell_wmi_descriptor ptdma ipmi_devintf rapl ipmi_msghandler acpi_power_meter pcspkr k10temp xfs libcrc32c sd_mod t10_pi mgag200 sg drm_kms_helper crct10dif_pclmul i2c_algo_bit ahci crc32_pclmul drm_shmem_helper libahci crc32c_intel drm i40e libata ghash_clmulni_intel tg3 ccp megaraid_sas sp5100_tco wmi dm_mirror dm_region_hash dm_log dm_mod fuse [ 200.869387] CPU: 91 PID: 13634 Comm: userfaultfd Kdump: loaded Not tainted 6.9.0-rc5+ #8 [ 200.877477] Hardware name: Dell Inc. PowerEdge R6525/0N7YGH, BIOS 2.7.3 03/30/2022 [ 200.885052] RIP: 0010:zap_pte_range+0x43d/0x660 [ 200.889595] Code: 83 fa 02 0f 86 44 01 00 00 83 f9 17 0f 84 e1 00 00 00 83 f9 1f 0f 84 d0 00 00 00 48 89 c6 48 c7 c7 00 e4 dd bb e8 73 a2 de ff <0f> 0b e9 44 fd ff ff 45 0f b6 44 24 20 41 f6 c0 f4 75 27 4c 89 ee [ 200.908348] RSP: 0018:ffffa18d2e6c37c8 EFLAGS: 00010246 [ 200.913584] RAX: 000000000000002a RBX: 00007f26d3600000 RCX: 0000000000000000 [ 200.920730] RDX: 0000000000000000 RSI: ffff93503f9a0bc0 RDI: ffff93503f9a0bc0 [ 200.927867] RBP: 00007f26d35cc000 R08: 0000000000000000 R09: ffffa18d2e6c3688 [ 200.935009] R10: ffffa18d2e6c3680 R11: ffffffffbc9de448 R12: ffffa18d2e6c39e8 [ 200.942149] R13: ffff92d1ebc15b50 R14: ffff93114e0cde60 R15: ffffa18d2e6c3928 [ 200.949291] FS: 0000000000000000(0000) GS:ffff93503f980000(0000) knlGS:0000000000000000 [ 200.957384] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 200.963140] CR2: 00007f26b1600658 CR3: 00000040905ba000 CR4: 0000000000350ef0 [ 200.970283] Call Trace: [ 200.972745] <TASK> [ 200.974862] ? __warn+0x7f/0x130 [ 200.978108] ? zap_pte_range+0x43d/0x660 [ 200.982044] ? report_bug+0x18a/0x1a0 [ 200.985720] ? handle_bug+0x3c/0x70 [ 200.989219] ? exc_invalid_op+0x14/0x70 [ 200.993068] ? asm_exc_invalid_op+0x16/0x20 [ 200.997265] ? zap_pte_range+0x43d/0x660 [ 201.001199] ? zap_pte_range+0x43d/0x660 [ 201.005134] zap_pmd_range.isra.0+0xf9/0x230 [ 201.009416] unmap_page_range+0x2d4/0x4a0 [ 201.013436] unmap_vmas+0xa8/0x180 [ 201.016854] exit_mmap+0xea/0x3b0 [ 201.020191] __mmput+0x43/0x120 [ 201.023342] exit_mm+0xb1/0x110 [ 201.026496] do_exit+0x270/0x4f0 [ 201.029739] do_group_exit+0x2c/0x80 [ 201.033326] get_signal+0x886/0x8b0 [ 201.036828] ? srso_return_thunk+0x5/0x5f [ 201.040848] arch_do_signal_or_restart+0x25/0x100 [ 201.045563] ? srso_return_thunk+0x5/0x5f [ 201.049583] ? vma_set_page_prot+0x5e/0xc0 [ 201.053692] ? srso_return_thunk+0x5/0x5f [ 201.057713] ? syscall_exit_work+0xff/0x130 [ 201.061908] syscall_exit_to_user_mode+0x1b3/0x200 [ 201.066712] do_syscall_64+0x87/0x160 [ 201.070387] ? srso_return_thunk+0x5/0x5f [ 201.074405] ? do_mmap+0x416/0x5f0 [ 201.077821] ? srso_return_thunk+0x5/0x5f [ 201.081840] ? rseq_get_rseq_cs+0x1d/0x240 [ 201.085950] ? srso_return_thunk+0x5/0x5f [ 201.089970] ? rseq_ip_fixup+0x6d/0x1d0 [ 201.093823] ? vm_mmap_pgoff+0x117/0x1a0 [ 201.097755] ? srso_return_thunk+0x5/0x5f [ 201.101776] ? srso_return_thunk+0x5/0x5f [ 201.105795] ? syscall_exit_to_user_mode+0x78/0x200 [ 201.110685] ? srso_return_thunk+0x5/0x5f [ 201.114706] ? do_syscall_64+0x87/0x160 [ 201.118557] ? srso_return_thunk+0x5/0x5f [ 201.122575] ? __count_memcg_events+0x49/0xb0 [ 201.126944] ? srso_return_thunk+0x5/0x5f [ 201.130967] ? srso_return_thunk+0x5/0x5f [ 201.134986] ? syscall_exit_work+0xff/0x130 [ 201.139184] ? srso_return_thunk+0x5/0x5f [ 201.143205] ? syscall_exit_to_user_mode+0x78/0x200 [ 201.148093] ? srso_return_thunk+0x5/0x5f [ 201.152114] ? do_syscall_64+0x87/0x160 [ 201.155960] ? srso_return_thunk+0x5/0x5f [ 201.159984] ? sched_clock_cpu+0xb/0x190 [ 201.163916] ? srso_return_thunk+0x5/0x5f [ 201.167939] ? irqtime_account_irq+0x40/0xc0 [ 201.172220] ? srso_return_thunk+0x5/0x5f [ 201.176243] ? srso_return_thunk+0x5/0x5f [ 201.180263] entry_SYSCALL_64_after_hwframe+0x76/0x7e [ 201.185326] RIP: 0033:0x7f26dfd0735b [ 201.188939] Code: Unable to access opcode bytes at 0x7f26dfd07331. [ 201.195128] RSP: 002b:00007fffce176868 EFLAGS: 00000206 ORIG_RAX: 000000000000000a [ 201.202700] RAX: fffffffffffffffc RBX: 00007f26dfe60000 RCX: 00007f26dfd0735b [ 201.209841] RDX: 0000000000000003 RSI: 0000000001000000 RDI: 00007f26af401000 [ 201.216983] RBP: 00007f26b0400640 R08: 00000000ffffffff R09: 0000000000000000 [ 201.224127] R10: ffffffffffffffc0 R11: 0000000000000206 R12: 0000000000000000 [ 201.231267] R13: 000000000040d320 R14: 0000000000000000 R15: 0000000000000000 [ 201.238413] </TASK> [ 201.240610] ---[ end trace 0000000000000000 ]--- [ 201.245250] unrecognized swap entry 0x7c00000000000001Quoted 2 lines
Likely we want "Fixes:" here.
Quoted 7 lines
This could be seen as a continuation of the problem
2ff559f31a5dRevert "userfaultfd: don't fail on unrecognized features" was trying to solve. However, this patch only checks to make sure we didnt ask for a feature outside the possible range of features. We are still missing a check to confirm the requested features are also configured on. So I guess the "Fixes" tag would be for this patch?914eedcb9ba0userfaultfd: don't fail on unrecognized features
Even though 914eedcb9ba0 was problematic but it might be innocent in this regard, as we miss such check even before that commit.
Perhaps the right Fixes should be when there will be flags that will only be supported conditionally, which should be the wp support. In that case, it could be:
But it's still complicated though even if we want to backport, as this specific check for userfault was added later... in:
00b151f21f39 ("mm/userfaultfd: fail uffd-wp registration if not supported")
In summary: I think we can stick with Fixes on e06f1e1dd499, but we don't copy stable. The major reason we don't copy stable here is not only about complexity of such backport, but also that there can be apps trying to pass in unsupported bits (even if the kernel didn't support it) but keep using MISSING mode only, then we shouldn't fail them easily after a stable upgrade. Just smells dangerous to backport.
When at it and repost, would you mind consolidate the two feature list check together? We had the previous check here:
if (uffdio_api.api != UFFD_API || (features & ~UFFD_API_FEATURES)) goto err_out;
If with this patch IIUC this previous check can be removed (still need to check UFFD_API).
Thanks,
Quoted 38 lines
Happy to get your input here!
Thanks in advance!
Quoted 31 lines
Quoted 13 lines
Signed-off-by Audra Mitchell <[email protected]>--- fs/userfaultfd.c | 5 +++++ 1 file changed, 5 insertions(+)fs/userfaultfd.c +0
-0@@ -2073,6 +2073,11 @@ static int userfaultfd_api(struct userfaultfd_ctx *ctx, uffdio_api.features &= ~UFFD_FEATURE_WP_UNPOPULATED ;fs/userfaultfd.c static int userfaultfd_api(struct userfaultfd_ctx *ctx, · line 2083
Quoted 10 lines
fs/userfaultfd.c +5
-0uffdio_api.features &= ~UFFD_FEATURE_WP_ASYNC ;#endif + + ret = -EINVAL ; +if (features & ~uffdio_api.features) + goto err_out; + uffdio_api.ioctls =UFFD_API_IOCTLS ; ret = -EFAULT ;if (copy_to_user (buf, &uffdio_api,sizeof (uffdio_api)))CCing Peter.
Signature · 3 lines
Cheers, David / dhildenb
Signature · 1 line
Peter Xu
Quoted 197 lines · Peter Xu
Quoted 124 lines · Audra Mitchell
Quoted 13 lines · David Hildenbrand
Quoted 3 lines · Audra Mitchell
Currently if we request a feature that is not set in the Kernel config we fail silently and return the available features. However, the documentation indicates we should return an EINVAL.
I assume you are referencing
"EINVAL The API version requested in the api field is not supported by this kernel, or the features field passed to the kernel includes feature bits that are not supported by the current kernel version."
and
Quoted 8 lines
"To enable userfaultfd features the application should set a bit corresponding to each feature it wants to enable in the features field. If the kernel supports all the requested features it will enable them. Otherwise it will zero out the returned
uffdio_apistructure and return EINVAL.
"in which case I agree.
Yep! I'm referencing the man page.
Quoted 7 lines
Quoted 5 lines
We need to fix this issue since we can end up with a Kernel warning should a program request the feature
UFFD_FEATURE_WP_UNPOPULATEDon a kernel with the config not set with this feature.Can you mention which exact one? Is it a WARN* or a
pr_warn()?Here is the kernel warning I get:
[ 200.803094] unrecognized swap entry 0x7c00000000000001 [ 200.808270] ------------[ cut here ]------------ [ 200.812896] WARNING: CPU: 91 PID: 13634 at mm/memory.c:1660 zap_pte_range+0x43d/0x660 [ 200.820738] Modules linked in: qrtr bridge stp llc rfkill sunrpc amd_atl intel_rapl_msr intel_rapl_common amd64_edac edac_mce_amd kvm_amd kvm ipmi_ssif acpi_ipmi i2c_piix4 ipmi_si wmi_bmof dcdbas dell_smbios dell_wmi_descriptor ptdma ipmi_devintf rapl ipmi_msghandler acpi_power_meter pcspkr k10temp xfs libcrc32c sd_mod t10_pi mgag200 sg drm_kms_helper crct10dif_pclmul i2c_algo_bit ahci crc32_pclmul drm_shmem_helper libahci crc32c_intel drm i40e libata ghash_clmulni_intel tg3 ccp megaraid_sas sp5100_tco wmi dm_mirror dm_region_hash dm_log dm_mod fuse [ 200.869387] CPU: 91 PID: 13634 Comm: userfaultfd Kdump: loaded Not tainted 6.9.0-rc5+ #8 [ 200.877477] Hardware name: Dell Inc. PowerEdge R6525/0N7YGH, BIOS 2.7.3 03/30/2022 [ 200.885052] RIP: 0010:zap_pte_range+0x43d/0x660 [ 200.889595] Code: 83 fa 02 0f 86 44 01 00 00 83 f9 17 0f 84 e1 00 00 00 83 f9 1f 0f 84 d0 00 00 00 48 89 c6 48 c7 c7 00 e4 dd bb e8 73 a2 de ff <0f> 0b e9 44 fd ff ff 45 0f b6 44 24 20 41 f6 c0 f4 75 27 4c 89 ee [ 200.908348] RSP: 0018:ffffa18d2e6c37c8 EFLAGS: 00010246 [ 200.913584] RAX: 000000000000002a RBX: 00007f26d3600000 RCX: 0000000000000000 [ 200.920730] RDX: 0000000000000000 RSI: ffff93503f9a0bc0 RDI: ffff93503f9a0bc0 [ 200.927867] RBP: 00007f26d35cc000 R08: 0000000000000000 R09: ffffa18d2e6c3688 [ 200.935009] R10: ffffa18d2e6c3680 R11: ffffffffbc9de448 R12: ffffa18d2e6c39e8 [ 200.942149] R13: ffff92d1ebc15b50 R14: ffff93114e0cde60 R15: ffffa18d2e6c3928 [ 200.949291] FS: 0000000000000000(0000) GS:ffff93503f980000(0000) knlGS:0000000000000000 [ 200.957384] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 200.963140] CR2: 00007f26b1600658 CR3: 00000040905ba000 CR4: 0000000000350ef0 [ 200.970283] Call Trace: [ 200.972745] <TASK> [ 200.974862] ? __warn+0x7f/0x130 [ 200.978108] ? zap_pte_range+0x43d/0x660 [ 200.982044] ? report_bug+0x18a/0x1a0 [ 200.985720] ? handle_bug+0x3c/0x70 [ 200.989219] ? exc_invalid_op+0x14/0x70 [ 200.993068] ? asm_exc_invalid_op+0x16/0x20 [ 200.997265] ? zap_pte_range+0x43d/0x660 [ 201.001199] ? zap_pte_range+0x43d/0x660 [ 201.005134] zap_pmd_range.isra.0+0xf9/0x230 [ 201.009416] unmap_page_range+0x2d4/0x4a0 [ 201.013436] unmap_vmas+0xa8/0x180 [ 201.016854] exit_mmap+0xea/0x3b0 [ 201.020191] __mmput+0x43/0x120 [ 201.023342] exit_mm+0xb1/0x110 [ 201.026496] do_exit+0x270/0x4f0 [ 201.029739] do_group_exit+0x2c/0x80 [ 201.033326] get_signal+0x886/0x8b0 [ 201.036828] ? srso_return_thunk+0x5/0x5f [ 201.040848] arch_do_signal_or_restart+0x25/0x100 [ 201.045563] ? srso_return_thunk+0x5/0x5f [ 201.049583] ? vma_set_page_prot+0x5e/0xc0 [ 201.053692] ? srso_return_thunk+0x5/0x5f [ 201.057713] ? syscall_exit_work+0xff/0x130 [ 201.061908] syscall_exit_to_user_mode+0x1b3/0x200 [ 201.066712] do_syscall_64+0x87/0x160 [ 201.070387] ? srso_return_thunk+0x5/0x5f [ 201.074405] ? do_mmap+0x416/0x5f0 [ 201.077821] ? srso_return_thunk+0x5/0x5f [ 201.081840] ? rseq_get_rseq_cs+0x1d/0x240 [ 201.085950] ? srso_return_thunk+0x5/0x5f [ 201.089970] ? rseq_ip_fixup+0x6d/0x1d0 [ 201.093823] ? vm_mmap_pgoff+0x117/0x1a0 [ 201.097755] ? srso_return_thunk+0x5/0x5f [ 201.101776] ? srso_return_thunk+0x5/0x5f [ 201.105795] ? syscall_exit_to_user_mode+0x78/0x200 [ 201.110685] ? srso_return_thunk+0x5/0x5f [ 201.114706] ? do_syscall_64+0x87/0x160 [ 201.118557] ? srso_return_thunk+0x5/0x5f [ 201.122575] ? __count_memcg_events+0x49/0xb0 [ 201.126944] ? srso_return_thunk+0x5/0x5f [ 201.130967] ? srso_return_thunk+0x5/0x5f [ 201.134986] ? syscall_exit_work+0xff/0x130 [ 201.139184] ? srso_return_thunk+0x5/0x5f [ 201.143205] ? syscall_exit_to_user_mode+0x78/0x200 [ 201.148093] ? srso_return_thunk+0x5/0x5f [ 201.152114] ? do_syscall_64+0x87/0x160 [ 201.155960] ? srso_return_thunk+0x5/0x5f [ 201.159984] ? sched_clock_cpu+0xb/0x190 [ 201.163916] ? srso_return_thunk+0x5/0x5f [ 201.167939] ? irqtime_account_irq+0x40/0xc0 [ 201.172220] ? srso_return_thunk+0x5/0x5f [ 201.176243] ? srso_return_thunk+0x5/0x5f [ 201.180263] entry_SYSCALL_64_after_hwframe+0x76/0x7e [ 201.185326] RIP: 0033:0x7f26dfd0735b [ 201.188939] Code: Unable to access opcode bytes at 0x7f26dfd07331. [ 201.195128] RSP: 002b:00007fffce176868 EFLAGS: 00000206 ORIG_RAX: 000000000000000a [ 201.202700] RAX: fffffffffffffffc RBX: 00007f26dfe60000 RCX: 00007f26dfd0735b [ 201.209841] RDX: 0000000000000003 RSI: 0000000001000000 RDI: 00007f26af401000 [ 201.216983] RBP: 00007f26b0400640 R08: 00000000ffffffff R09: 0000000000000000 [ 201.224127] R10: ffffffffffffffc0 R11: 0000000000000206 R12: 0000000000000000 [ 201.231267] R13: 000000000040d320 R14: 0000000000000000 R15: 0000000000000000 [ 201.238413] </TASK> [ 201.240610] ---[ end trace 0000000000000000 ]--- [ 201.245250] unrecognized swap entry 0x7c00000000000001Quoted 2 lines
Likely we want "Fixes:" here.
Quoted 7 lines
This could be seen as a continuation of the problem
2ff559f31a5dRevert "userfaultfd: don't fail on unrecognized features" was trying to solve. However, this patch only checks to make sure we didnt ask for a feature outside the possible range of features. We are still missing a check to confirm the requested features are also configured on. So I guess the "Fixes" tag would be for this patch?914eedcb9ba0userfaultfd: don't fail on unrecognized featuresEven though
914eedcb9ba0was problematic but it might be innocent in this regard, as we miss such check even before that commit.Perhaps the right Fixes should be when there will be flags that will only be supported conditionally, which should be the wp support. In that case, it could be:
Fixes e06f1e1dd499 ("userfaultfd: wp: enabled write protection in userfaultfd API")But it's still complicated though even if we want to backport, as this specific check for userfault was added later... in:
00b151f21f39("mm/userfaultfd: fail uffd-wp registration if not supported")In summary: I think we can stick with Fixes on
e06f1e1dd499, but we don't copy stable. The major reason we don't copy stable here is not only about complexity of such backport, but also that there can be apps trying to pass in unsupported bits (even if the kernel didn't support it) but keep using MISSING mode only, then we shouldn't fail them easily after a stable upgrade. Just smells dangerous to backport.When at it and repost, would you mind consolidate the two feature list check together? We had the previous check here:
if (uffdio_api.api != UFFD_API || (features & ~UFFD_API_FEATURES)) goto err_out;If with this patch IIUC this previous check can be removed (still need to check
UFFD_API).Thanks,
Quoted 32 lines
Happy to get your input here!
Thanks in advance!
Quoted 26 lines
Quoted 13 lines
Signed-off-by Audra Mitchell <[email protected]>--- fs/userfaultfd.c | 5 +++++ 1 file changed, 5 insertions(+)fs/userfaultfd.c +0
-0@@ -2073,6 +2073,11 @@ static int userfaultfd_api(struct userfaultfd_ctx *ctx, uffdio_api.features &= ~UFFD_FEATURE_WP_UNPOPULATED ;fs/userfaultfd.c static int userfaultfd_api(struct userfaultfd_ctx *ctx, · line 2083
Quoted 10 lines
fs/userfaultfd.c +5
-0uffdio_api.features &= ~UFFD_FEATURE_WP_ASYNC ;#endif + + ret = -EINVAL ; +if (features & ~uffdio_api.features) + goto err_out; + uffdio_api.ioctls =UFFD_API_IOCTLS ; ret = -EFAULT ;if (copy_to_user (buf, &uffdio_api,sizeof (uffdio_api)))CCing Peter.
Quoted 9 lines
Quoted 6 lines
Quoted 5 lines
Signature · 3 lines
Cheers, David / dhildenbSignature · 1 line
Peter Xu
I didn't forget about this, but I did find another test case that is going to break with this change. Some other things landed on my plate/schedule so I'll be returning to this in a week or two. I'll make sure I take your feedback when I submit v2. Thanks again!
Quoted 1 line
Quoted 21 lines · Audra Mitchell
Quoted 20 lines · David Hildenbrand
Quoted 3 lines · Audra Mitchell
Currently if we request a feature that is not set in the Kernel config we fail silently and return the available features. However, the documentation indicates we should return an EINVAL.
I assume you are referencing
"EINVAL The API version requested in the api field is not supported by this kernel, or the features field passed to the kernel includes feature bits that are not supported by the current kernel version."
and
"To enable userfaultfd features the application should set a bit corresponding to each feature it wants to enable in the features field. If the kernel supports all the requested features it will enable them. Otherwise it will zero out the returned
uffdio_apistructure and return EINVAL.
"
Quoted 3 lines
Quoted 1 line
in which case I agree.
Yep! I'm referencing the man page.
Might be worth just quoting the man page :)
Quoted 8 lines
Quoted 5 lines
Quoted 3 lines
We need to fix this issue since we can end up with a Kernel warning should a program request the feature
UFFD_FEATURE_WP_UNPOPULATEDon a kernel with the config not set with this feature.Can you mention which exact one? Is it a WARN* or a
pr_warn()?Here is the kernel warning I get:
Thanks, it makes sense to add the first couple of lines of below to the patch description.
Quoted 97 lines
[ 200.803094] unrecognized swap entry 0x7c00000000000001 [ 200.808270] ------------[ cut here ]------------ [ 200.812896] WARNING: CPU: 91 PID: 13634 at mm/memory.c:1660 zap_pte_range+0x43d/0x660 [ 200.820738] Modules linked in: qrtr bridge stp llc rfkill sunrpc amd_atl intel_rapl_msr intel_rapl_common amd64_edac edac_mce_amd kvm_amd kvm ipmi_ssif acpi_ipmi i2c_piix4 ipmi_si wmi_bmof dcdbas dell_smbios dell_wmi_descriptor ptdma ipmi_devintf rapl ipmi_msghandler acpi_power_meter pcspkr k10temp xfs libcrc32c sd_mod t10_pi mgag200 sg drm_kms_helper crct10dif_pclmul i2c_algo_bit ahci crc32_pclmul drm_shmem_helper libahci crc32c_intel drm i40e libata ghash_clmulni_intel tg3 ccp megaraid_sas sp5100_tco wmi dm_mirror dm_region_hash dm_log dm_mod fuse [ 200.869387] CPU: 91 PID: 13634 Comm: userfaultfd Kdump: loaded Not tainted 6.9.0-rc5+ #8 [ 200.877477] Hardware name: Dell Inc. PowerEdge R6525/0N7YGH, BIOS 2.7.3 03/30/2022 [ 200.885052] RIP: 0010:zap_pte_range+0x43d/0x660 [ 200.889595] Code: 83 fa 02 0f 86 44 01 00 00 83 f9 17 0f 84 e1 00 00 00 83 f9 1f 0f 84 d0 00 00 00 48 89 c6 48 c7 c7 00 e4 dd bb e8 73 a2 de ff <0f> 0b e9 44 fd ff ff 45 0f b6 44 24 20 41 f6 c0 f4 75 27 4c 89 ee [ 200.908348] RSP: 0018:ffffa18d2e6c37c8 EFLAGS: 00010246 [ 200.913584] RAX: 000000000000002a RBX: 00007f26d3600000 RCX: 0000000000000000 [ 200.920730] RDX: 0000000000000000 RSI: ffff93503f9a0bc0 RDI: ffff93503f9a0bc0 [ 200.927867] RBP: 00007f26d35cc000 R08: 0000000000000000 R09: ffffa18d2e6c3688 [ 200.935009] R10: ffffa18d2e6c3680 R11: ffffffffbc9de448 R12: ffffa18d2e6c39e8 [ 200.942149] R13: ffff92d1ebc15b50 R14: ffff93114e0cde60 R15: ffffa18d2e6c3928 [ 200.949291] FS: 0000000000000000(0000) GS:ffff93503f980000(0000) knlGS:0000000000000000 [ 200.957384] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 200.963140] CR2: 00007f26b1600658 CR3: 00000040905ba000 CR4: 0000000000350ef0 [ 200.970283] Call Trace: [ 200.972745] <TASK> [ 200.974862] ? __warn+0x7f/0x130 [ 200.978108] ? zap_pte_range+0x43d/0x660 [ 200.982044] ? report_bug+0x18a/0x1a0 [ 200.985720] ? handle_bug+0x3c/0x70 [ 200.989219] ? exc_invalid_op+0x14/0x70 [ 200.993068] ? asm_exc_invalid_op+0x16/0x20 [ 200.997265] ? zap_pte_range+0x43d/0x660 [ 201.001199] ? zap_pte_range+0x43d/0x660 [ 201.005134] zap_pmd_range.isra.0+0xf9/0x230 [ 201.009416] unmap_page_range+0x2d4/0x4a0 [ 201.013436] unmap_vmas+0xa8/0x180 [ 201.016854] exit_mmap+0xea/0x3b0 [ 201.020191] __mmput+0x43/0x120 [ 201.023342] exit_mm+0xb1/0x110 [ 201.026496] do_exit+0x270/0x4f0 [ 201.029739] do_group_exit+0x2c/0x80 [ 201.033326] get_signal+0x886/0x8b0 [ 201.036828] ? srso_return_thunk+0x5/0x5f [ 201.040848] arch_do_signal_or_restart+0x25/0x100 [ 201.045563] ? srso_return_thunk+0x5/0x5f [ 201.049583] ? vma_set_page_prot+0x5e/0xc0 [ 201.053692] ? srso_return_thunk+0x5/0x5f [ 201.057713] ? syscall_exit_work+0xff/0x130 [ 201.061908] syscall_exit_to_user_mode+0x1b3/0x200 [ 201.066712] do_syscall_64+0x87/0x160 [ 201.070387] ? srso_return_thunk+0x5/0x5f [ 201.074405] ? do_mmap+0x416/0x5f0 [ 201.077821] ? srso_return_thunk+0x5/0x5f [ 201.081840] ? rseq_get_rseq_cs+0x1d/0x240 [ 201.085950] ? srso_return_thunk+0x5/0x5f [ 201.089970] ? rseq_ip_fixup+0x6d/0x1d0 [ 201.093823] ? vm_mmap_pgoff+0x117/0x1a0 [ 201.097755] ? srso_return_thunk+0x5/0x5f [ 201.101776] ? srso_return_thunk+0x5/0x5f [ 201.105795] ? syscall_exit_to_user_mode+0x78/0x200 [ 201.110685] ? srso_return_thunk+0x5/0x5f [ 201.114706] ? do_syscall_64+0x87/0x160 [ 201.118557] ? srso_return_thunk+0x5/0x5f [ 201.122575] ? __count_memcg_events+0x49/0xb0 [ 201.126944] ? srso_return_thunk+0x5/0x5f [ 201.130967] ? srso_return_thunk+0x5/0x5f [ 201.134986] ? syscall_exit_work+0xff/0x130 [ 201.139184] ? srso_return_thunk+0x5/0x5f [ 201.143205] ? syscall_exit_to_user_mode+0x78/0x200 [ 201.148093] ? srso_return_thunk+0x5/0x5f [ 201.152114] ? do_syscall_64+0x87/0x160 [ 201.155960] ? srso_return_thunk+0x5/0x5f [ 201.159984] ? sched_clock_cpu+0xb/0x190 [ 201.163916] ? srso_return_thunk+0x5/0x5f [ 201.167939] ? irqtime_account_irq+0x40/0xc0 [ 201.172220] ? srso_return_thunk+0x5/0x5f [ 201.176243] ? srso_return_thunk+0x5/0x5f [ 201.180263] entry_SYSCALL_64_after_hwframe+0x76/0x7e [ 201.185326] RIP: 0033:0x7f26dfd0735b [ 201.188939] Code: Unable to access opcode bytes at 0x7f26dfd07331. [ 201.195128] RSP: 002b:00007fffce176868 EFLAGS: 00000206 ORIG_RAX: 000000000000000a [ 201.202700] RAX: fffffffffffffffc RBX: 00007f26dfe60000 RCX: 00007f26dfd0735b [ 201.209841] RDX: 0000000000000003 RSI: 0000000001000000 RDI: 00007f26af401000 [ 201.216983] RBP: 00007f26b0400640 R08: 00000000ffffffff R09: 0000000000000000 [ 201.224127] R10: ffffffffffffffc0 R11: 0000000000000206 R12: 0000000000000000 [ 201.231267] R13: 000000000040d320 R14: 0000000000000000 R15: 0000000000000000 [ 201.238413] </TASK> [ 201.240610] ---[ end trace 0000000000000000 ]--- [ 201.245250] unrecognized swap entry 0x7c00000000000001Quoted 2 lines
Likely we want "Fixes:" here.
This could be seen as a continuation of the problem
2ff559f31a5dRevert "userfaultfd: don't fail on unrecognized features" was trying to solve. However, this patch only checks to make sure we didnt ask for a feature outside the possible range of features. We are still missing a check to confirm the requested features are also configured on. So I guess the "Fixes" tag would be for this patch?914eedcb9ba0userfaultfd: don't fail on unrecognized features
Quoted 1 line
Happy to get your input here!
I agree with Peters summary :)
Signature · 3 lines
Cheers, David / dhildenb