search

rhboot / shim

UEFI shim loader
Other
848 stars 290 forks source link

Fix: Keep BOOT CSV be Unicode format file #513

Closed Rtoax closed 1 year ago

Rtoax commented 1 year ago

In CentOS and AlmaLinux, the CSV file:

00000000  ff fe 73 00 68 00 69 00  6d 00 69 00 61 00 33 00  |..s.h.i.m.i.a.3.| 
00000010  32 00 2e 00 65 00 66 00  69 00 2c 00 43 00 65 00  |2...e.f.i.,.C.e.|

It's unicode file marked by 'ff fe' characters, however, iconv -t UCS-2LE not provided:

00000000  73 00 68 00 69 00 6d 00  61 00 61 00 36 00 34 00  |s.h.i.m.a.a.6.4.| 
00000010  2e 00 65 00 66 00 69 00  2c 00 41 00 6c 00 6d 00  |..e.f.i.,.A.l.m.|
julian-klode commented 1 year ago

I'm not sure what your issue is, but the file is supposed to be UCS-2 little-endian like everything else in UEFI. Adding this change would break the parser at the very least as it does not expect a BOM, and then you also need to deal with transcoding back to UCS-2 any additional characters only available in Unicode in the fallback code.

Rtoax commented 1 year ago

@julian-klode

Thanks for your reply, I just found the CSV file format is not as same as CentOS, there is no 'ff fe' as first two char:

$ make install-deps OSLABEL=AlmaLinux
$ file BOOTAA64.CSV 
BOOTAA64.CSV: data
# Hexdump
00000000  73 00 68 00 69 00 6d 00  61 00 61 00 36 00 34 00  |s.h.i.m.a.a.6.4.| 
00000010  2e 00 65 00 66 00 69 00  2c 00 41 00 6c 00 6d 00  |..e.f.i.,.A.l.m.|

However, CentOS's BOOTAA64.CSV:

$ file BOOTAA64.CSV 
BOOTAA64.CSV: Little-endian UTF-16 Unicode text
# Hexdump
00000000  ff fe 73 00 68 00 69 00  6d 00 69 00 61 00 33 00  |..s.h.i.m.i.a.3.| 
00000010  32 00 2e 00 65 00 66 00  69 00 2c 00 43 00 65 00  |2...e.f.i.,.C.e.|

because first two char 'ff fe', if it's doesn't matters, please ignore this PR.