brintos

brintos / linux-shallow public Read only

0
0
Text · 18.5 KiB · 9072f21 Raw
478 lines · plain
1.. SPDX-License-Identifier: GPL-2.0-only2 3============4UAPI Checker5============6 7The UAPI checker (``scripts/check-uapi.sh``) is a shell script which8checks UAPI header files for userspace backwards-compatibility across9the git tree.10 11Options12=======13 14This section will describe the options with which ``check-uapi.sh``15can be run.16 17Usage::18 19    check-uapi.sh [-b BASE_REF] [-p PAST_REF] [-j N] [-l ERROR_LOG] [-i] [-q] [-v]20 21Available options::22 23    -b BASE_REF    Base git reference to use for comparison. If unspecified or empty,24                   will use any dirty changes in tree to UAPI files. If there are no25                   dirty changes, HEAD will be used.26    -p PAST_REF    Compare BASE_REF to PAST_REF (e.g. -p v6.1). If unspecified or empty,27                   will use BASE_REF^1. Must be an ancestor of BASE_REF. Only headers28                   that exist on PAST_REF will be checked for compatibility.29    -j JOBS        Number of checks to run in parallel (default: number of CPU cores).30    -l ERROR_LOG   Write error log to file (default: no error log is generated).31    -i             Ignore ambiguous changes that may or may not break UAPI compatibility.32    -q             Quiet operation.33    -v             Verbose operation (print more information about each header being checked).34 35Environmental args::36 37    ABIDIFF  Custom path to abidiff binary38    CC       C compiler (default is "gcc")39    ARCH     Target architecture of C compiler (default is host arch)40 41Exit codes::42 43    0) Success44    1) ABI difference detected45    2) Prerequisite not met46 47Examples48========49 50Basic Usage51-----------52 53First, let's try making a change to a UAPI header file that obviously54won't break userspace::55 56    cat << 'EOF' | patch -l -p157    --- a/include/uapi/linux/acct.h58    +++ b/include/uapi/linux/acct.h59    @@ -21,7 +21,9 @@60     #include <asm/param.h>61     #include <asm/byteorder.h>62 63    -/*64    +#define FOO65    +66    +/*67      *  comp_t is a 16-bit "floating" point number with a 3-bit base 868      *  exponent and a 13-bit fraction.69      *  comp2_t is 24-bit with 5-bit base 2 exponent and 20 bit fraction70    diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h71    EOF72 73Now, let's use the script to validate::74 75    % ./scripts/check-uapi.sh76    Installing user-facing UAPI headers from dirty tree... OK77    Installing user-facing UAPI headers from HEAD... OK78    Checking changes to UAPI headers between HEAD and dirty tree...79    All 912 UAPI headers compatible with x86 appear to be backwards compatible80 81Let's add another change that *might* break userspace::82 83    cat << 'EOF' | patch -l -p184    --- a/include/uapi/linux/bpf.h85    +++ b/include/uapi/linux/bpf.h86    @@ -74,7 +74,7 @@ struct bpf_insn {87            __u8    dst_reg:4;      /* dest register */88            __u8    src_reg:4;      /* source register */89            __s16   off;            /* signed offset */90    -       __s32   imm;            /* signed immediate constant */91    +       __u32   imm;            /* unsigned immediate constant */92     };93 94     /* Key of an a BPF_MAP_TYPE_LPM_TRIE entry */95    EOF96 97The script will catch this::98 99    % ./scripts/check-uapi.sh100    Installing user-facing UAPI headers from dirty tree... OK101    Installing user-facing UAPI headers from HEAD... OK102    Checking changes to UAPI headers between HEAD and dirty tree...103    ==== ABI differences detected in include/linux/bpf.h from HEAD -> dirty tree ====104        [C] 'struct bpf_insn' changed:105          type size hasn't changed106          1 data member change:107            type of '__s32 imm' changed:108              typedef name changed from __s32 to __u32 at int-ll64.h:27:1109              underlying type 'int' changed:110                type name changed from 'int' to 'unsigned int'111                type size hasn't changed112    ==================================================================================113 114    error - 1/912 UAPI headers compatible with x86 appear _not_ to be backwards compatible115 116In this case, the script is reporting the type change because it could117break a userspace program that passes in a negative number. Now, let's118say you know that no userspace program could possibly be using a negative119value in ``imm``, so changing to an unsigned type there shouldn't hurt120anything. You can pass the ``-i`` flag to the script to ignore changes121in which the userspace backwards compatibility is ambiguous::122 123    % ./scripts/check-uapi.sh -i124    Installing user-facing UAPI headers from dirty tree... OK125    Installing user-facing UAPI headers from HEAD... OK126    Checking changes to UAPI headers between HEAD and dirty tree...127    All 912 UAPI headers compatible with x86 appear to be backwards compatible128 129Now, let's make a similar change that *will* break userspace::130 131    cat << 'EOF' | patch -l -p1132    --- a/include/uapi/linux/bpf.h133    +++ b/include/uapi/linux/bpf.h134    @@ -71,8 +71,8 @@ enum {135 136     struct bpf_insn {137            __u8    code;           /* opcode */138    -       __u8    dst_reg:4;      /* dest register */139            __u8    src_reg:4;      /* source register */140    +       __u8    dst_reg:4;      /* dest register */141            __s16   off;            /* signed offset */142            __s32   imm;            /* signed immediate constant */143     };144    EOF145 146Since we're re-ordering an existing struct member, there's no ambiguity,147and the script will report the breakage even if you pass ``-i``::148 149    % ./scripts/check-uapi.sh -i150    Installing user-facing UAPI headers from dirty tree... OK151    Installing user-facing UAPI headers from HEAD... OK152    Checking changes to UAPI headers between HEAD and dirty tree...153    ==== ABI differences detected in include/linux/bpf.h from HEAD -> dirty tree ====154        [C] 'struct bpf_insn' changed:155          type size hasn't changed156          2 data member changes:157            '__u8 dst_reg' offset changed from 8 to 12 (in bits) (by +4 bits)158            '__u8 src_reg' offset changed from 12 to 8 (in bits) (by -4 bits)159    ==================================================================================160 161    error - 1/912 UAPI headers compatible with x86 appear _not_ to be backwards compatible162 163Let's commit the breaking change, then commit the innocuous change::164 165    % git commit -m 'Breaking UAPI change' include/uapi/linux/bpf.h166    [detached HEAD f758e574663a] Breaking UAPI change167     1 file changed, 1 insertion(+), 1 deletion(-)168    % git commit -m 'Innocuous UAPI change' include/uapi/linux/acct.h169    [detached HEAD 2e87df769081] Innocuous UAPI change170     1 file changed, 3 insertions(+), 1 deletion(-)171 172Now, let's run the script again with no arguments::173 174    % ./scripts/check-uapi.sh175    Installing user-facing UAPI headers from HEAD... OK176    Installing user-facing UAPI headers from HEAD^1... OK177    Checking changes to UAPI headers between HEAD^1 and HEAD...178    All 912 UAPI headers compatible with x86 appear to be backwards compatible179 180It doesn't catch any breaking change because, by default, it only181compares ``HEAD`` to ``HEAD^1``. The breaking change was committed on182``HEAD~2``. If we wanted the search scope to go back further, we'd have to183use the ``-p`` option to pass a different past reference. In this case,184let's pass ``-p HEAD~2`` to the script so it checks UAPI changes between185``HEAD~2`` and ``HEAD``::186 187    % ./scripts/check-uapi.sh -p HEAD~2188    Installing user-facing UAPI headers from HEAD... OK189    Installing user-facing UAPI headers from HEAD~2... OK190    Checking changes to UAPI headers between HEAD~2 and HEAD...191    ==== ABI differences detected in include/linux/bpf.h from HEAD~2 -> HEAD ====192        [C] 'struct bpf_insn' changed:193          type size hasn't changed194          2 data member changes:195            '__u8 dst_reg' offset changed from 8 to 12 (in bits) (by +4 bits)196            '__u8 src_reg' offset changed from 12 to 8 (in bits) (by -4 bits)197    ==============================================================================198 199    error - 1/912 UAPI headers compatible with x86 appear _not_ to be backwards compatible200 201Alternatively, we could have also run with ``-b HEAD~``. This would set the202base reference to ``HEAD~`` so then the script would compare it to ``HEAD~^1``.203 204Architecture-specific Headers205-----------------------------206 207Consider this change::208 209    cat << 'EOF' | patch -l -p1210    --- a/arch/arm64/include/uapi/asm/sigcontext.h211    +++ b/arch/arm64/include/uapi/asm/sigcontext.h212    @@ -70,6 +70,7 @@ struct sigcontext {213     struct _aarch64_ctx {214            __u32 magic;215            __u32 size;216    +       __u32 new_var;217     };218 219     #define FPSIMD_MAGIC   0x46508001220    EOF221 222This is a change to an arm64-specific UAPI header file. In this example, I'm223running the script from an x86 machine with an x86 compiler, so, by default,224the script only checks x86-compatible UAPI header files::225 226    % ./scripts/check-uapi.sh227    Installing user-facing UAPI headers from dirty tree... OK228    Installing user-facing UAPI headers from HEAD... OK229    No changes to UAPI headers were applied between HEAD and dirty tree230 231With an x86 compiler, we can't check header files in ``arch/arm64``, so the232script doesn't even try.233 234If we want to check the header file, we'll have to use an arm64 compiler and235set ``ARCH`` accordingly::236 237    % CC=aarch64-linux-gnu-gcc ARCH=arm64 ./scripts/check-uapi.sh238    Installing user-facing UAPI headers from dirty tree... OK239    Installing user-facing UAPI headers from HEAD... OK240    Checking changes to UAPI headers between HEAD and dirty tree...241    ==== ABI differences detected in include/asm/sigcontext.h from HEAD -> dirty tree ====242        [C] 'struct _aarch64_ctx' changed:243          type size changed from 64 to 96 (in bits)244          1 data member insertion:245            '__u32 new_var', at offset 64 (in bits) at sigcontext.h:73:1246        -- snip --247        [C] 'struct zt_context' changed:248          type size changed from 128 to 160 (in bits)249          2 data member changes (1 filtered):250            '__u16 nregs' offset changed from 64 to 96 (in bits) (by +32 bits)251            '__u16 __reserved[3]' offset changed from 80 to 112 (in bits) (by +32 bits)252    =======================================================================================253 254    error - 1/884 UAPI headers compatible with arm64 appear _not_ to be backwards compatible255 256We can see with ``ARCH`` and ``CC`` set properly for the file, the ABI257change is reported properly. Also notice that the total number of UAPI258header files checked by the script changes. This is because the number259of headers installed for arm64 platforms is different than x86.260 261Cross-Dependency Breakages262--------------------------263 264Consider this change::265 266    cat << 'EOF' | patch -l -p1267    --- a/include/uapi/linux/types.h268    +++ b/include/uapi/linux/types.h269    @@ -52,7 +52,7 @@ typedef __u32 __bitwise __wsum;270     #define __aligned_be64 __be64 __attribute__((aligned(8)))271     #define __aligned_le64 __le64 __attribute__((aligned(8)))272 273    -typedef unsigned __bitwise __poll_t;274    +typedef unsigned short __bitwise __poll_t;275 276     #endif /*  __ASSEMBLY__ */277     #endif /* _UAPI_LINUX_TYPES_H */278    EOF279 280Here, we're changing a ``typedef`` in ``types.h``. This doesn't break281a UAPI in ``types.h``, but other UAPIs in the tree may break due to282this change::283 284    % ./scripts/check-uapi.sh285    Installing user-facing UAPI headers from dirty tree... OK286    Installing user-facing UAPI headers from HEAD... OK287    Checking changes to UAPI headers between HEAD and dirty tree...288    ==== ABI differences detected in include/linux/eventpoll.h from HEAD -> dirty tree ====289        [C] 'struct epoll_event' changed:290          type size changed from 96 to 80 (in bits)291          2 data member changes:292            type of '__poll_t events' changed:293              underlying type 'unsigned int' changed:294                type name changed from 'unsigned int' to 'unsigned short int'295                type size changed from 32 to 16 (in bits)296            '__u64 data' offset changed from 32 to 16 (in bits) (by -16 bits)297    ========================================================================================298    include/linux/eventpoll.h did not change between HEAD and dirty tree...299    It's possible a change to one of the headers it includes caused this error:300    #include <linux/fcntl.h>301    #include <linux/types.h>302 303Note that the script noticed the failing header file did not change,304so it assumes one of its includes must have caused the breakage. Indeed,305we can see ``linux/types.h`` is used from ``eventpoll.h``.306 307UAPI Header Removals308--------------------309 310Consider this change::311 312    cat << 'EOF' | patch -l -p1313    diff --git a/include/uapi/asm-generic/Kbuild b/include/uapi/asm-generic/Kbuild314    index ebb180aac74e..a9c88b0a8b3b 100644315    --- a/include/uapi/asm-generic/Kbuild316    +++ b/include/uapi/asm-generic/Kbuild317    @@ -31,6 +31,6 @@ mandatory-y += stat.h318     mandatory-y += statfs.h319     mandatory-y += swab.h320     mandatory-y += termbits.h321    -mandatory-y += termios.h322    +#mandatory-y += termios.h323     mandatory-y += types.h324     mandatory-y += unistd.h325    EOF326 327This script removes a UAPI header file from the install list. Let's run328the script::329 330    % ./scripts/check-uapi.sh331    Installing user-facing UAPI headers from dirty tree... OK332    Installing user-facing UAPI headers from HEAD... OK333    Checking changes to UAPI headers between HEAD and dirty tree...334    ==== UAPI header include/asm/termios.h was removed between HEAD and dirty tree ====335 336    error - 1/912 UAPI headers compatible with x86 appear _not_ to be backwards compatible337 338Removing a UAPI header is considered a breaking change, and the script339will flag it as such.340 341Checking Historic UAPI Compatibility342------------------------------------343 344You can use the ``-b`` and ``-p`` options to examine different chunks of your345git tree. For example, to check all changed UAPI header files between tags346v6.0 and v6.1, you'd run::347 348    % ./scripts/check-uapi.sh -b v6.1 -p v6.0349    Installing user-facing UAPI headers from v6.1... OK350    Installing user-facing UAPI headers from v6.0... OK351    Checking changes to UAPI headers between v6.0 and v6.1...352 353    --- snip ---354    error - 37/907 UAPI headers compatible with x86 appear _not_ to be backwards compatible355 356Note: Before v5.3, a header file needed by the script is not present,357so the script is unable to check changes before then.358 359You'll notice that the script detected many UAPI changes that are not360backwards compatible. Knowing that kernel UAPIs are supposed to be stable361forever, this is an alarming result. This brings us to the next section:362caveats.363 364Caveats365=======366 367The UAPI checker makes no assumptions about the author's intention, so some368types of changes may be flagged even though they intentionally break UAPI.369 370Removals For Refactoring or Deprecation371---------------------------------------372 373Sometimes drivers for very old hardware are removed, such as in this example::374 375    % ./scripts/check-uapi.sh -b ba47652ba655376    Installing user-facing UAPI headers from ba47652ba655... OK377    Installing user-facing UAPI headers from ba47652ba655^1... OK378    Checking changes to UAPI headers between ba47652ba655^1 and ba47652ba655...379    ==== UAPI header include/linux/meye.h was removed between ba47652ba655^1 and ba47652ba655 ====380 381    error - 1/910 UAPI headers compatible with x86 appear _not_ to be backwards compatible382 383The script will always flag removals (even if they're intentional).384 385Struct Expansions386-----------------387 388Depending on how a structure is handled in kernelspace, a change which389expands a struct could be non-breaking.390 391If a struct is used as the argument to an ioctl, then the kernel driver392must be able to handle ioctl commands of any size. Beyond that, you need393to be careful when copying data from the user. Say, for example, that394``struct foo`` is changed like this::395 396    struct foo {397        __u64 a; /* added in version 1 */398    +   __u32 b; /* added in version 2 */399    +   __u32 c; /* added in version 2 */400    }401 402By default, the script will flag this kind of change for further review::403 404    [C] 'struct foo' changed:405      type size changed from 64 to 128 (in bits)406      2 data member insertions:407        '__u32 b', at offset 64 (in bits)408        '__u32 c', at offset 96 (in bits)409 410However, it is possible that this change was made safely.411 412If a userspace program was built with version 1, it will think413``sizeof(struct foo)`` is 8. That size will be encoded in the414ioctl value that gets sent to the kernel. If the kernel is built415with version 2, it will think the ``sizeof(struct foo)`` is 16.416 417The kernel can use the ``_IOC_SIZE`` macro to get the size encoded418in the ioctl code that the user passed in and then use419``copy_struct_from_user()`` to safely copy the value::420 421    int handle_ioctl(unsigned long cmd, unsigned long arg)422    {423        switch _IOC_NR(cmd) {424        0x01: {425            struct foo my_cmd;  /* size 16 in the kernel */426 427            ret = copy_struct_from_user(&my_cmd, arg, sizeof(struct foo), _IOC_SIZE(cmd));428            ...429 430``copy_struct_from_user`` will zero the struct in the kernel and then copy431only the bytes passed in from the user (leaving new members zeroized).432If the user passed in a larger struct, the extra members are ignored.433 434If you know this situation is accounted for in the kernel code, you can435pass ``-i`` to the script, and struct expansions like this will be ignored.436 437Flex Array Migration438--------------------439 440While the script handles expansion into an existing flex array, it does441still flag initial migration to flex arrays from 1-element fake flex442arrays. For example::443 444    struct foo {445          __u32 x;446    -     __u32 flex[1]; /* fake flex */447    +     __u32 flex[];  /* real flex */448    };449 450This change would be flagged by the script::451 452    [C] 'struct foo' changed:453      type size changed from 64 to 32 (in bits)454      1 data member change:455        type of '__u32 flex[1]' changed:456          type name changed from '__u32[1]' to '__u32[]'457          array type size changed from 32 to 'unknown'458          array type subrange 1 changed length from 1 to 'unknown'459 460At this time, there's no way to filter these types of changes, so be461aware of this possible false positive.462 463Summary464-------465 466While many types of false positives are filtered out by the script,467it's possible there are some cases where the script flags a change468which does not break UAPI. It's also possible a change which *does*469break userspace would not be flagged by this script. While the script470has been run on much of the kernel history, there could still be corner471cases that are not accounted for.472 473The intention is for this script to be used as a quick check for474maintainers or automated tooling, not as the end-all authority on475patch compatibility. It's best to remember: use your best judgment476(and ideally a unit test in userspace) to make sure your UAPI changes477are backwards-compatible!478