mm/page_alloc: Require pahole v1.22 to cope with zero-sized struct pagesets

This patch replaces mm-page_alloc-convert-per-cpu-list-protection-to-local_lock-fix.patch in Andrew's tree.

Michal Suchanek reported the following problem with linux-next

  [    0.000000] Linux version 5.13.0-rc2-next-20210519-1.g3455ff8-vanilla (geeko@buildhost) (gcc (SUSE Linux) 10.3.0, GNU ld (GNU Binutils; openSUSE Tumbleweed) 2.36.1.20210326-3) #1 SMP Wed May 19 10:05:10 UTC 2021 (3455ff8)
  [    0.000000] Command line: BOOT_IMAGE=/boot/vmlinuz-5.13.0-rc2-next-20210519-1.g3455ff8-vanilla root=UUID=ec42c33e-a2c2-4c61-afcc-93e9527 8f687 plymouth.enable=0 resume=/dev/disk/by-uuid/f1fe4560-a801-4faf-a638-834c407027c7 mitigations=auto earlyprintk initcall_debug nomodeset earlycon ignore_loglevel console=ttyS0,115200
...
  [   26.093364] calling  tracing_set_default_clock+0x0/0x62 @ 1
  [   26.098937] initcall tracing_set_default_clock+0x0/0x62 returned 0 after 0 usecs
  [   26.106330] calling  acpi_gpio_handle_deferred_request_irqs+0x0/0x7c @ 1
  [   26.113033] initcall acpi_gpio_handle_deferred_request_irqs+0x0/0x7c returned 0 after 3 usecs
  [   26.121559] calling  clk_disable_unused+0x0/0x102 @ 1
  [   26.126620] initcall clk_disable_unused+0x0/0x102 returned 0 after 0 usecs
  [   26.133491] calling  regulator_init_complete+0x0/0x25 @ 1
  [   26.138890] initcall regulator_init_complete+0x0/0x25 returned 0 after 0 usecs
  [   26.147816] Freeing unused decrypted memory: 2036K
  [   26.153682] Freeing unused kernel image (initmem) memory: 2308K
  [   26.165776] Write protecting the kernel read-only data: 26624k
  [   26.173067] Freeing unused kernel image (text/rodata gap) memory: 2036K
  [   26.180416] Freeing unused kernel image (rodata/data gap) memory: 1184K
  [   26.187031] Run /init as init process
  [   26.190693]   with arguments:
  [   26.193661]     /init
  [   26.195933]   with environment:
  [   26.199079]     HOME=/
  [   26.201444]     TERM=linux
  [   26.204152]     BOOT_IMAGE=/boot/vmlinuz-5.13.0-rc2-next-20210519-1.g3455ff8-vanilla
  [   26.254154] BPF:      type_id=35503 offset=178440 size=4
  [   26.259125] BPF:
  [   26.261054] BPF:Invalid offset
  [   26.264119] BPF:
  [   26.264119]
  [   26.267437] failed to validate module [efivarfs] BTF: -22

Andrii Nakryiko bisected the problem to the commit "mm/page_alloc: convert per-cpu list protection to local_lock" currently staged in mmotm. In his own words

The immediate problem is two different definitions of numa_node per-cpu variable. They both are at the same offset within .data..percpu ELF section, they both have the same name, but one of them is marked as static and another as global. And one is int variable, while another is struct pagesets. I'll look some more tomorrow, but adding Jiri and Arnaldo for visibility.

[110907] DATASEC '.data..percpu' size=178904 vlen=303
...

        type_id=27753 offset=163976 size=4 (VAR 'numa_node')
        type_id=27754 offset=163976 size=4 (VAR 'numa_node')

[27753] VAR 'numa_node' type_id=27556, linkage=static [27754] VAR 'numa_node' type_id=20, linkage=global

[20] INT 'int' size=4 bits_offset=0 nr_bits=32 encoding=SIGNED

[27556] STRUCT 'pagesets' size=0 vlen=1

        'lock' type_id=507 bits_offset=0

[506] STRUCT '(anon)' size=0 vlen=0
[507] TYPEDEF 'local_lock_t' type_id=506

The patch in question introduces a zero-sized per-cpu struct and while this is not wrong, versions of pahole prior to 1.22 get confused during BTF generation with two separate variables occupying the same address.

This patch adds a requirement for pahole 1.22 before setting DEBUG_INFO_BTF. While pahole 1.22 does not exist yet, a fix is in the pahole git tree as ("btf_encoder: fix and complete filtering out zero-sized per-CPU variables").

Reported-by Michal Suchanek <[email protected]>

Reported-by Hritik Vijay <[email protected]>

Debugged-by Andrii Nakryiko <[email protected]>

Signed-off-by Mel Gorman <[email protected]>

---

 lib/Kconfig.debug | 4 ++++
 1 file changed, 4 insertions(+)

lib/Kconfig.debug +4 -0

index 678c13967580..825be101767e 100644
@@ -301,10 +301,14 @@ config DEBUG_INFO_DWARF5
 
 endchoice # "DWARF version"
 
+config PAHOLE_HAS_ZEROSIZE_PERCPU_SUPPORT
+       def_bool $(success, test `$(PAHOLE) --version | sed -E 's/v([0-9]+)\.([0-9]+)/\1\2/'` -ge "122")
+
 config DEBUG_INFO_BTF
        bool "Generate BTF typeinfo"
        depends on !DEBUG_INFO_SPLIT && !DEBUG_INFO_REDUCED
        depends on !GCC_PLUGIN_RANDSTRUCT || COMPILE_TEST
+       depends on PAHOLE_HAS_ZEROSIZE_PERCPU_SUPPORT
        help
          Generate deduplicated BTF type information from DWARF debug info.
          Turning this on expects presence of pahole tool, which will convert

Quoted 36 lines

Andrii Nakryiko bisected the problem to the commit "mm/page_alloc: convert per-cpu list protection to local_lock" currently staged in mmotm. In his own words

The immediate problem is two different definitions of numa_node per-cpu variable. They both are at the same offset within .data..percpu ELF section, they both have the same name, but one of them is marked as static and another as global. And one is int variable, while another is struct pagesets. I'll look some more tomorrow, but adding Jiri and Arnaldo for visibility.

[110907] DATASEC '.data..percpu' size=178904 vlen=303
...

        type_id=27753 offset=163976 size=4 (VAR 'numa_node')
        type_id=27754 offset=163976 size=4 (VAR 'numa_node')

[27753] VAR 'numa_node' type_id=27556, linkage=static [27754] VAR 'numa_node' type_id=20, linkage=global

[20] INT 'int' size=4 bits_offset=0 nr_bits=32 encoding=SIGNED

[27556] STRUCT 'pagesets' size=0 vlen=1

        'lock' type_id=507 bits_offset=0

[506] STRUCT '(anon)' size=0 vlen=0
[507] TYPEDEF 'local_lock_t' type_id=506

The patch in question introduces a zero-sized per-cpu struct and while this is not wrong, versions of pahole prior to 1.22 get confused during BTF generation with two separate variables occupying the same address.

This patch adds a requirement for pahole 1.22 before setting DEBUG_INFO_BTF. While pahole 1.22 does not exist yet, a fix is in the pahole git tree as ("btf_encoder: fix and complete filtering out zero-sized per-CPU variables").

Quoted 5 lines

Reported-by Michal Suchanek <[email protected]>

Reported-by Hritik Vijay <[email protected]>

Debugged-by Andrii Nakryiko <[email protected]>

Signed-off-by Mel Gorman <[email protected]>

---

I still think that v1 ([0]) is a more appropriate temporary solution until pahole 1.22 is released and widely packaged. Suddenly raising the minimum version to 1.22, which is not even released even, is a pretty big compatibility concern for all the users that rely on CONFIG_DEBUG_INFO_BTF.

Just a few days ago pahole 1.16 worked fine and
here we suddenly (and silently due to how Kconfig functions) raise that to a version that doesn't exist. That's going to break workflows for a lot of people.

I'm asking to have that ugly work-around to ensure sizeof(struct pagesets) > 0 as a temporary solution only.

Quoted 42 lines · Andrii Nakryiko

Quoted 36 lines

Andrii Nakryiko bisected the problem to the commit "mm/page_alloc: convert per-cpu list protection to local_lock" currently staged in mmotm. In his own words

The immediate problem is two different definitions of numa_node per-cpu variable. They both are at the same offset within .data..percpu ELF section, they both have the same name, but one of them is marked as static and another as global. And one is int variable, while another is struct pagesets. I'll look some more tomorrow, but adding Jiri and Arnaldo for visibility.

[110907] DATASEC '.data..percpu' size=178904 vlen=303
...

        type_id=27753 offset=163976 size=4 (VAR 'numa_node')
        type_id=27754 offset=163976 size=4 (VAR 'numa_node')

[27753] VAR 'numa_node' type_id=27556, linkage=static [27754] VAR 'numa_node' type_id=20, linkage=global

[20] INT 'int' size=4 bits_offset=0 nr_bits=32 encoding=SIGNED

[27556] STRUCT 'pagesets' size=0 vlen=1

        'lock' type_id=507 bits_offset=0

[506] STRUCT '(anon)' size=0 vlen=0
[507] TYPEDEF 'local_lock_t' type_id=506

The patch in question introduces a zero-sized per-cpu struct and while this is not wrong, versions of pahole prior to 1.22 get confused during BTF generation with two separate variables occupying the same address.

This patch adds a requirement for pahole 1.22 before setting DEBUG_INFO_BTF. While pahole 1.22 does not exist yet, a fix is in the pahole git tree as ("btf_encoder: fix and complete filtering out zero-sized per-CPU variables").

Quoted 5 lines

Reported-by Michal Suchanek <[email protected]>

Reported-by Hritik Vijay <[email protected]>

Debugged-by Andrii Nakryiko <[email protected]>

Signed-off-by Mel Gorman <[email protected]>

---

Quoted 5 lines

I still think that v1 ([0]) is a more appropriate temporary solution until pahole 1.22 is released and widely packaged. Suddenly raising the minimum version to 1.22, which is not even released even, is a pretty big compatibility concern for all the users that rely on CONFIG_DEBUG_INFO_BTF.

On the flip side, we have a situation where a build tool (pahole) has a problem whereby correct code does not result in a working kernel. It's not that dissimilar to preventing the kernel being built on an old compiler. While I accept it's unfortunate, Christoph had a point where introducing workarounds in the kernel could lead to a prolification of workarounds for pahole or other reasons that are potentially tricky to revert as long as distributions exist that do not ship with a sufficiently reason package.

Quoted 5 lines

Just a few days ago pahole 1.16 worked fine and
here we suddenly (and silently due to how Kconfig functions) raise that to a version that doesn't exist. That's going to break workflows for a lot of people.

People do have a workaround though. For the system building the kernel, they can patch pahole and revert the check so a bootable kernel can be built. It's not convenient but it is manageable and pahole has until 5.13 releases to release a v1.22. The downsides for the alternative -- a non-booting kernel are much more severe.

Quoted 2 lines

I'm asking to have that ugly work-around to ensure sizeof(struct pagesets) > 0 as a temporary solution only.

Another temporary solution is to locally build pahole and either revert the check or fake the 1.22 release number with the self-built pahole.

--
Mel Gorman
SUSE Labs

Quoted 71 lines · Mel Gorman

Quoted 42 lines · Andrii Nakryiko

Quoted 36 lines

Andrii Nakryiko bisected the problem to the commit "mm/page_alloc: convert per-cpu list protection to local_lock" currently staged in mmotm. In his own words

The immediate problem is two different definitions of numa_node per-cpu variable. They both are at the same offset within .data..percpu ELF section, they both have the same name, but one of them is marked as static and another as global. And one is int variable, while another is struct pagesets. I'll look some more tomorrow, but adding Jiri and Arnaldo for visibility.

[110907] DATASEC '.data..percpu' size=178904 vlen=303
...

        type_id=27753 offset=163976 size=4 (VAR 'numa_node')
        type_id=27754 offset=163976 size=4 (VAR 'numa_node')

[27753] VAR 'numa_node' type_id=27556, linkage=static [27754] VAR 'numa_node' type_id=20, linkage=global

[20] INT 'int' size=4 bits_offset=0 nr_bits=32 encoding=SIGNED

[27556] STRUCT 'pagesets' size=0 vlen=1

        'lock' type_id=507 bits_offset=0

[506] STRUCT '(anon)' size=0 vlen=0
[507] TYPEDEF 'local_lock_t' type_id=506

The patch in question introduces a zero-sized per-cpu struct and while this is not wrong, versions of pahole prior to 1.22 get confused during BTF generation with two separate variables occupying the same address.

This patch adds a requirement for pahole 1.22 before setting DEBUG_INFO_BTF. While pahole 1.22 does not exist yet, a fix is in the pahole git tree as ("btf_encoder: fix and complete filtering out zero-sized per-CPU variables").

Quoted 5 lines

Reported-by Michal Suchanek <[email protected]>

Reported-by Hritik Vijay <[email protected]>

Debugged-by Andrii Nakryiko <[email protected]>

Signed-off-by Mel Gorman <[email protected]>

---

Quoted 5 lines

I still think that v1 ([0]) is a more appropriate temporary solution until pahole 1.22 is released and widely packaged. Suddenly raising the minimum version to 1.22, which is not even released even, is a pretty big compatibility concern for all the users that rely on CONFIG_DEBUG_INFO_BTF.

On the flip side, we have a situation where a build tool (pahole) has a problem whereby correct code does not result in a working kernel. It's not that dissimilar to preventing the kernel being built on an old compiler. While I accept it's unfortunate, Christoph had a point where introducing workarounds in the kernel could lead to a prolification of workarounds for pahole or other reasons that are potentially tricky to revert as long as distributions exist that do not ship with a sufficiently reason package.

Quoted 5 lines

Just a few days ago pahole 1.16 worked fine and
here we suddenly (and silently due to how Kconfig functions) raise that to a version that doesn't exist. That's going to break workflows for a lot of people.

People do have a workaround though. For the system building the kernel, they can patch pahole and revert the check so a bootable kernel can be built. It's not convenient but it is manageable and pahole has until 5.13 releases to release a v1.22. The downsides for the alternative -- a non-booting kernel are much more severe.

Quoted 6 lines

Quoted 2 lines

I'm asking to have that ugly work-around to ensure sizeof(struct pagesets) > 0 as a temporary solution only.

Another temporary solution is to locally build pahole and either revert the check or fake the 1.22 release number with the self-built pahole.

Well, luckily it seems we anticipated issues like that and added --skip_encoding_btf_vars argument, which I completely forgot about and just accidentally came across reviewing Arnaldo's latest pahole patch. I think that one is a much better solution, as then it will impact only those that explicitly relies on availability of BTF for per-CPU variables, which is a subset of all possible uses for kernel BTF. Sent a patch ([0]), please take a look.

[0] https://lore.kernel.org/linux-mm/[email protected]/T/#u

Quoted 87 lines · Andrii Nakryiko

Quoted 71 lines · Mel Gorman

Quoted 42 lines · Andrii Nakryiko

Quoted 36 lines

Andrii Nakryiko bisected the problem to the commit "mm/page_alloc: convert per-cpu list protection to local_lock" currently staged in mmotm. In his own words

The immediate problem is two different definitions of numa_node per-cpu variable. They both are at the same offset within .data..percpu ELF section, they both have the same name, but one of them is marked as static and another as global. And one is int variable, while another is struct pagesets. I'll look some more tomorrow, but adding Jiri and Arnaldo for visibility.

[110907] DATASEC '.data..percpu' size=178904 vlen=303
...

        type_id=27753 offset=163976 size=4 (VAR 'numa_node')
        type_id=27754 offset=163976 size=4 (VAR 'numa_node')

[27753] VAR 'numa_node' type_id=27556, linkage=static [27754] VAR 'numa_node' type_id=20, linkage=global

[20] INT 'int' size=4 bits_offset=0 nr_bits=32 encoding=SIGNED

[27556] STRUCT 'pagesets' size=0 vlen=1

        'lock' type_id=507 bits_offset=0

[506] STRUCT '(anon)' size=0 vlen=0
[507] TYPEDEF 'local_lock_t' type_id=506

The patch in question introduces a zero-sized per-cpu struct and while this is not wrong, versions of pahole prior to 1.22 get confused during BTF generation with two separate variables occupying the same address.

This patch adds a requirement for pahole 1.22 before setting DEBUG_INFO_BTF. While pahole 1.22 does not exist yet, a fix is in the pahole git tree as ("btf_encoder: fix and complete filtering out zero-sized per-CPU variables").

Quoted 5 lines

Reported-by Michal Suchanek <[email protected]>

Reported-by Hritik Vijay <[email protected]>

Debugged-by Andrii Nakryiko <[email protected]>

Signed-off-by Mel Gorman <[email protected]>

---

Quoted 5 lines

I still think that v1 ([0]) is a more appropriate temporary solution until pahole 1.22 is released and widely packaged. Suddenly raising the minimum version to 1.22, which is not even released even, is a pretty big compatibility concern for all the users that rely on CONFIG_DEBUG_INFO_BTF.

On the flip side, we have a situation where a build tool (pahole) has a problem whereby correct code does not result in a working kernel. It's not that dissimilar to preventing the kernel being built on an old compiler. While I accept it's unfortunate, Christoph had a point where introducing workarounds in the kernel could lead to a prolification of workarounds for pahole or other reasons that are potentially tricky to revert as long as distributions exist that do not ship with a sufficiently reason package.

Quoted 5 lines

Just a few days ago pahole 1.16 worked fine and
here we suddenly (and silently due to how Kconfig functions) raise that to a version that doesn't exist. That's going to break workflows for a lot of people.

People do have a workaround though. For the system building the kernel, they can patch pahole and revert the check so a bootable kernel can be built. It's not convenient but it is manageable and pahole has until 5.13 releases to release a v1.22. The downsides for the alternative -- a non-booting kernel are much more severe.

Quoted 6 lines

Quoted 2 lines

I'm asking to have that ugly work-around to ensure sizeof(struct pagesets) > 0 as a temporary solution only.

Another temporary solution is to locally build pahole and either revert the check or fake the 1.22 release number with the self-built pahole.

Well, luckily it seems we anticipated issues like that and added --skip_encoding_btf_vars argument, which I completely forgot about and just accidentally came across reviewing Arnaldo's latest pahole patch. I think that one is a much better solution, as then it will impact only those that explicitly relies on availability of BTF for per-CPU variables, which is a subset of all possible uses for kernel BTF. Sent a patch ([0]), please take a look.

Quoted 1 line

[0] https://lore.kernel.org/linux-mm/[email protected]/T/#u

I'm happy to have this patch used as an alternative to forcing 1.22 to be the minimum version of pahole required.

Quoted 8 lines

Quoted 8 lines

Well, luckily it seems we anticipated issues like that and added --skip_encoding_btf_vars argument, which I completely forgot about and just accidentally came across reviewing Arnaldo's latest pahole patch. I think that one is a much better solution, as then it will impact only those that explicitly relies on availability of BTF for per-CPU variables, which is a subset of all possible uses for kernel BTF. Sent a patch ([0]), please take a look.

Quoted 4 lines

Quoted 1 line

[0] https://lore.kernel.org/linux-mm/[email protected]/T/#u

I'm happy to have this patch used as an alternative to forcing 1.22 to be the minimum version of pahole required.

Is pahole 1.22 available already?