---
version 0.12.4:
- Workaround for broken OSS_GETVERSION on FreeBSD, part two (Juergen Lock)
- oss: fix fragment setting (malc)
- oss: issue OSS_GETVERSION ioctl only when needed (malc)
- oss: refactor code around policy setting (malc)
- oss: workaround for cases when OSS_GETVERSION is not defined (malc)
- block: Free iovec arrays allocated by multiwrite_merge() (Stefan Hajnoczi)
- lsi: fix segfault in lsi_command_complete (Gerd Hoffmann)
- lsi: pass lsi_request to lsi_reselect (Gerd Hoffmann)
- lsi: move dma_len+dma_buf into lsi_request (Gerd Hoffmann)
- lsi: move current_dev into lsi_request (Gerd Hoffmann)
- lsi: have lsi_request for the whole life time of the request. (Gerd Hoffmann)
- lsi: use QTAILQ for lsi_queue (Gerd Hoffmann)
- tcp/mips: Change TCG_AREG0 (fp -> s0) (Stefan Weil)
- sh_pci: fix memory and I/O access (Aurelien Jarno)
- Fix incoming migration with iothread (Marcelo Tosatti)
- Fix SIGFPE for vnc display of width/height = 1 (Chris Webb)
- net: remove broken net_set_boot_mask() boot device validation (Eduardo Habkost)
- qcow2: Remove request from in-flight list after error (Kevin Wolf)
- qcow2: Don't ignore immediate read/write failures (Kevin Wolf)
- block: Fix multiwrite memory leak in error case (Kevin Wolf)
- block: Fix error code in multiwrite for immediate failures (Kevin Wolf)
- block: Fix multiwrite error handling (Kevin Wolf)
- scsi-disk: fix buffer overflow (Gerd Hoffmann)
- qcow2: Rewrite alloc_refcount_block/grow_refcount_table (Kevin Wolf)
- qcow2: Factor next_refcount_table_size out (Kevin Wolf)
- block: avoid creating too large iovecs in multiwrite_merge (Christoph Hellwig)
- json-parser: Fix segfault on malformed input (Kevin Wolf)
- linux-user: switch default ppc64 CPU to 970fx from 970 (Aurelien Jarno)
- target-sh4: MMU: fix store queue addresses (Aurelien Jarno)
- target-sh4: MMU: fix ITLB priviledge check (Aurelien Jarno)
- target-sh4: MMU: fix mem_idx computation (Aurelien Jarno)
- sh7750: handle MMUCR TI bit (Aurelien Jarno)
- UHCI spurious interrut fix (Paul Brook)
- tcg/mips: fix branch offset during retranslation (Aurelien Jarno)
- tcg/arm: correctly save/restore registers in prologue/epilogue (Aurelien Jarno)
- workaround for cmd646 bmdma register access while no dma is active (Igor V. Kovalenko)
- Fix corner case in chardev udp: parameter (Jan Kiszka)
- Don't set default monitor when there is a mux'ed one (Jan Kiszka)
- spelling typo (compatibilty) in hw/fw_cfg.c (Vagrant Cascadian)
- fdc: fix drive property handling. (Gerd Hoffmann)
- target-i386: fix commit c22549204a6edc431e8e4358e61bd56386ff6957 (TeLeMan)
- target-i386: fix SIB decoding with index = 4 (Aurelien Jarno)
- Fix segfault with ram_size > 4095M without kvm (Ryan Harper)
- target-i386: Fix long jumps/calls in long mode with REX.W set (malc)
- target-i386: fix lddqu SSE instruction (Aurelien Jarno)
- qemu-char.c: drop debug printfs from qemu_chr_parse_compat (Jan Kiszka)
- fix undefined shifts by >32 (Paolo Bonzini)
- Fix qemu -net user,hostfwd= example (Aurelien Jarno)
---
Also add a patch for e1000.c NIC to make wm(4) work on NetBSD guest:
---
* Reset EEPROM internal state on CS rise edge,
not SK rise edge during CS is low, per CS pin description in
the FM93C06 EEPROM manual:
"... a rising edge on this signal is required to reset the
internal state-machine to accept a new cycle ..."
Intel's em driver seems to explicitly raise and lower SK output
after CS is negated in em_standby_eeprom() so many other OSes
that use Intel's driver don't have this problem, but I can't find
articles that say the MICROWIRE or EEPROM spec requires such sequence.
* It isn't a good idea to use memset(,0,) to clear EEPROM state values,
so explicitly clear status members and preserve the previous EECD values
---
This patch is also filed as upstream QEMU bugs #581737.
While here, several tweaks in Makefile to appease pkglint.
