Would this make sense for -stable? Impact is modest: fix a rather obscure feature of x86 targets on bigendian hosts, and improve error messages around it). On the other hand, the patches look pretty safe to me.
Markus Armbruster <arm...@redhat.com> writes: > Better error messages, a bit of code cleanup, and a big endian fix. > > Not addressed: qemu_uuid_parse() sets an SMBIOS field by side effect. > Gross! > > Testing: > * Verify error messages improve for > -smbios gaga > -smbios file=gaga > -smbios type=42 > -smbios type=1,uuid=gaga > -smbios type=0,release=gaga > * Verify SMBIOS table remains unchanged > -smbios type=0,vendor=me,version=42,date=today,release=1.2 -smbios > type=1,manufacturer=me,product=crap,version=6,serial=77,uuid=988bc9dd-0986-440f-ac24-cf9626c5aa88,sku=888,family=flintstones > by sticking > qemu_hexdump(smbios_entries, stdout, "SMBIOS", smbios_entries_len); > into smbios_get_table() > > v2: Address "Hawkeye" Laszlo's review > * 1-3/7 unchanged > * Drop 4/7 because it's buggy, and the fixed version isn't worthwhile > * Spelling fix in commit message of 5/7 > * Correct scanf format in 5-6/7 > > Markus Armbruster (6): > error-report.h: Supply missing include > log.h: Supply missing includes > smbios: Convert to error_report() > smbios: Clean up smbios_add_field() parameters > smbios: Fix -smbios type=0,release=... for big endian hosts > smbios: Check R in -smbios type=0,release=R parses okay > > arch_init.c | 3 +-- > hw/i386/smbios.c | 57 > ++++++++++++++++++++++++--------------------- > include/hw/i386/smbios.h | 2 +- > include/qemu/error-report.h | 1 + > include/qemu/log.h | 3 +++ > 5 files changed, 37 insertions(+), 29 deletions(-)
