diff options
| author | David Benjamin <davidben@google.com> | 2019-01-29 05:12:15 +0000 |
|---|---|---|
| committer | Kurt Roeckx <kurt@roeckx.be> | 2019-02-17 23:39:51 +0100 |
| commit | c0e8e5007ba5234d4d448e82a1567e0c4467e629 (patch) | |
| tree | 257257043dcf4b59cca63acde207333d8ddbac26 /crypto/perlasm | |
| parent | 8f58ede09572dcc6a7e6c01280dd348240199568 (diff) | |
Fix some CFI issues in x86_64 assembly
The add/double shortcut in ecp_nistz256-x86_64.pl left one instruction
point that did not unwind, and the "slow" path in AES_cbc_encrypt was
not annotated correctly. For the latter, add
.cfi_{remember,restore}_state support to perlasm.
Next, fill in a bunch of functions that are missing no-op .cfi_startproc
and .cfi_endproc blocks. libunwind cannot unwind those stack frames
otherwise.
Finally, work around a bug in libunwind by not encoding rflags. (rflags
isn't a callee-saved register, so there's not much need to annotate it
anyway.)
These were found as part of ABI testing work in BoringSSL.
Reviewed-by: Richard Levitte <levitte@openssl.org>
GH: #8109
Diffstat (limited to 'crypto/perlasm')
| -rwxr-xr-x | crypto/perlasm/x86_64-xlate.pl | 9 |
1 files changed, 9 insertions, 0 deletions
diff --git a/crypto/perlasm/x86_64-xlate.pl b/crypto/perlasm/x86_64-xlate.pl index 5e5d6cc0b1..080fc8d881 100755 --- a/crypto/perlasm/x86_64-xlate.pl +++ b/crypto/perlasm/x86_64-xlate.pl @@ -541,6 +541,7 @@ my %globals; ); my ($cfa_reg, $cfa_rsp); + my @cfa_stack; # [us]leb128 format is variable-length integer representation base # 2^128, with most significant bit of each byte being 0 denoting @@ -688,6 +689,14 @@ my %globals; cfa_expression($$line))); last; }; + /remember_state/ + && do { push @cfa_stack, [$cfa_reg, $cfa_rsp]; + last; + }; + /restore_state/ + && do { ($cfa_reg, $cfa_rsp) = @{pop @cfa_stack}; + last; + }; } $self->{value} = ".cfi_$dir\t$$line" if ($dir); |