| Message ID | 1370847839-5594-1-git-send-email-pl@kamp.de |
|---|---|
| State | New |
| Headers | show |
please ignore this one. Peter On 10.06.2013 09:03, Peter Lieven wrote: > migration relies on the target memory to be zeroed out since commit > f1c72795 (migration: do not sent zero pages in bulk stage). however, > there is a subtle case where this breaks migration. if for some reason > a page is zero at the source but not at the destination the destination > memory is corrupted. > > this was reported to break migration on pseries and also other platforms > might be affected. to ultimatively make sure the destination memory is zero > at the destination check for it on negotiation of ram blocks. > > note: the better fix for this would be to pass a flag to the machine > init functions of all architectures to indicate that the machine is a > migration target and then avoid copying ram images etc. to physical > ram in this case. but this would require a lot of code to be changed > and reviewed. > > Signed-off-by: Peter Lieven <pl@kamp.de> > --- > arch_init.c | 10 ++++++++++ > 1 file changed, 10 insertions(+) > > diff --git a/arch_init.c b/arch_init.c > index 5d32ecf..458bf8c 100644 > --- a/arch_init.c > +++ b/arch_init.c > @@ -799,6 +799,8 @@ static int ram_load(QEMUFile *f, void *opaque, int version_id) > while (total_ram_bytes) { > RAMBlock *block; > uint8_t len; > + void *base; > + ram_addr_t offset; > > len = qemu_get_byte(f); > qemu_get_buffer(f, (uint8_t *)id, len); > @@ -822,6 +824,14 @@ static int ram_load(QEMUFile *f, void *opaque, int version_id) > goto done; > } > > + base = memory_region_get_ram_ptr(block->mr); > + for (offset = 0; offset < block->length; > + offset += TARGET_PAGE_SIZE) { > + if (!is_zero_page(base + offset)) { > + memset(base + offset, 0x00, TARGET_PAGE_SIZE); > + } > + } > + > total_ram_bytes -= length; > } > }
diff --git a/arch_init.c b/arch_init.c index 5d32ecf..458bf8c 100644 --- a/arch_init.c +++ b/arch_init.c @@ -799,6 +799,8 @@ static int ram_load(QEMUFile *f, void *opaque, int version_id) while (total_ram_bytes) { RAMBlock *block; uint8_t len; + void *base; + ram_addr_t offset; len = qemu_get_byte(f); qemu_get_buffer(f, (uint8_t *)id, len); @@ -822,6 +824,14 @@ static int ram_load(QEMUFile *f, void *opaque, int version_id) goto done; } + base = memory_region_get_ram_ptr(block->mr); + for (offset = 0; offset < block->length; + offset += TARGET_PAGE_SIZE) { + if (!is_zero_page(base + offset)) { + memset(base + offset, 0x00, TARGET_PAGE_SIZE); + } + } + total_ram_bytes -= length; } }
migration relies on the target memory to be zeroed out since commit f1c72795 (migration: do not sent zero pages in bulk stage). however, there is a subtle case where this breaks migration. if for some reason a page is zero at the source but not at the destination the destination memory is corrupted. this was reported to break migration on pseries and also other platforms might be affected. to ultimatively make sure the destination memory is zero at the destination check for it on negotiation of ram blocks. note: the better fix for this would be to pass a flag to the machine init functions of all architectures to indicate that the machine is a migration target and then avoid copying ram images etc. to physical ram in this case. but this would require a lot of code to be changed and reviewed. Signed-off-by: Peter Lieven <pl@kamp.de> --- arch_init.c | 10 ++++++++++ 1 file changed, 10 insertions(+)