CVE-2022-49658

In the Linux kernel, the following vulnerability has been resolved:

bpf: Fix insufficient bounds propagation from adjustscalarminmaxvals

Kuee reported a corner case where the tnum becomes constant after the call to __regboundoffset(), but the register's bounds are not, that is, its min bounds are still not equal to the register's max bounds.

This in turn allows to leak pointers through turning a pointer register as is into an unknown scalar via adjustptrminmaxvals().

Before:

func#0 @0 0: R1=ctx(off=0,imm=0,umax=0,varoff=(0x0; 0x0)) R10=fp(off=0,imm=0,umax=0,varoff=(0x0; 0x0)) 0: (b7) r0 = 1 ; R0w=scalar(imm=1,umin=1,umax=1,varoff=(0x1; 0x0)) 1: (b7) r3 = 0 ; R3w=scalar(imm=0,umax=0,varoff=(0x0; 0x0)) 2: (87) r3 = -r3 ; R3w=scalar() 3: (87) r3 = -r3 ; R3w=scalar() 4: (47) r3 |= 32767 ; R3w=scalar(smin=-9223372036854743041,umin=32767,varoff=(0x7fff; 0xffffffffffff8000),s32min=-2147450881) 5: (75) if r3 s>= 0x0 goto pc+1 ; R3w=scalar(umin=9223372036854808575,varoff=(0x8000000000007fff; 0x7fffffffffff8000),s32min=-2147450881,u32_min=32767) 6: (95) exit

from 5 to 7: R0=scalar(imm=1,umin=1,umax=1,varoff=(0x1; 0x0)) R1=ctx(off=0,imm=0,umax=0,varoff=(0x0; 0x0)) R3=scalar(umin=32767,umax=9223372036854775807,varoff=(0x7fff; 0x7fffffffffff8000),s32min=-2147450881) R10=fp(off=0,imm=0,umax=0,varoff=(0x0; 0x0)) 7: (d5) if r3 s<= 0x8000 goto pc+1 ; R3=scalar(umin=32769,umax=9223372036854775807,varoff=(0x7fff; 0x7fffffffffff8000),s32min=-2147450881,u32min=32767) 8: (95) exit

from 7 to 9: R0=scalar(imm=1,umin=1,umax=1,varoff=(0x1; 0x0)) R1=ctx(off=0,imm=0,umax=0,varoff=(0x0; 0x0)) R3=scalar(umin=32767,umax=32768,varoff=(0x7fff; 0x8000)) R10=fp(off=0,imm=0,umax=0,varoff=(0x0; 0x0)) 9: (07) r3 += -32767 ; R3w=scalar(imm=0,umax=1,varoff=(0x0; 0x0)) <--- [*] 10: (95) exit

What can be seen here is that R3=scalar(umin=32767,umax=32768,varoff=(0x7fff; 0x8000)) after the operation R3 += -32767 results in a 'malformed' constant, that is, R3w=scalar(imm=0,umax=1,varoff=(0x0; 0x0)). Intersecting with varoff has not been done at that point via __updateregbounds(), which would have improved the umax to be equal to umin.

Refactor the tnum <> min/max bounds information flow into a regboundssync() helper and use it consistently everywhere. After the fix, bounds have been corrected to R3w=scalar(imm=0,umax=0,varoff=(0x0; 0x0)) and thus the register is regarded as a 'proper' constant scalar of 0.

After:

func#0 @0 0: R1=ctx(off=0,imm=0,umax=0,varoff=(0x0; 0x0)) R10=fp(off=0,imm=0,umax=0,varoff=(0x0; 0x0)) 0: (b7) r0 = 1 ; R0w=scalar(imm=1,umin=1,umax=1,varoff=(0x1; 0x0)) 1: (b7) r3 = 0 ; R3w=scalar(imm=0,umax=0,varoff=(0x0; 0x0)) 2: (87) r3 = -r3 ; R3w=scalar() 3: (87) r3 = -r3 ; R3w=scalar() 4: (47) r3 |= 32767 ; R3w=scalar(smin=-9223372036854743041,umin=32767,varoff=(0x7fff; 0xffffffffffff8000),s32min=-2147450881) 5: (75) if r3 s>= 0x0 goto pc+1 ; R3w=scalar(umin=9223372036854808575,varoff=(0x8000000000007fff; 0x7fffffffffff8000),s32min=-2147450881,u32_min=32767) 6: (95) exit

from 5 to 7: R0=scalar(imm=1,umin=1,umax=1,varoff=(0x1; 0x0)) R1=ctx(off=0,imm=0,umax=0,varoff=(0x0; 0x0)) R3=scalar(umin=32767,umax=9223372036854775807,varoff=(0x7fff; 0x7fffffffffff8000),s32min=-2147450881) R10=fp(off=0,imm=0,umax=0,varoff=(0x0; 0x0)) 7: (d5) if r3 s<= 0x8000 goto pc+1 ; R3=scalar(umin=32769,umax=9223372036854775807,varoff=(0x7fff; 0x7fffffffffff8000),s32min=-2147450881,u32min=32767) 8: (95) exit

from 7 to 9: R0=scalar(imm=1,umin=1,umax=1,varoff=(0x1; 0x0)) R1=ctx(off=0,imm=0,umax=0,varoff=(0x0; 0x0)) R3=scalar(umin=32767,umax=32768,var_off=(0x7fff; 0x8000)) R10=fp(off=0 ---truncated---