All isapc machines must have 32-bit CPUs and so the RAM split logic can be hardcoded
accordingly.
Signed-off-by: Mark Cave-Ayland <mark.caveayland@nutanix.com>
---
hw/i386/pc_piix.c | 58 ++++-------------------------------------------
1 file changed, 4 insertions(+), 54 deletions(-)
diff --git a/hw/i386/pc_piix.c b/hw/i386/pc_piix.c
index 816124c027..fc94937ad4 100644
--- a/hw/i386/pc_piix.c
+++ b/hw/i386/pc_piix.c
@@ -444,69 +444,19 @@ static void pc_init_isa(MachineState *machine)
GSIState *gsi_state;
MemoryRegion *ram_memory;
MemoryRegion *rom_memory = system_memory;
- ram_addr_t lowmem;
uint64_t hole64_size = 0;
/*
- * Calculate ram split, for memory below and above 4G. It's a bit
- * complicated for backward compatibility reasons ...
- *
- * - Traditional split is 3.5G (lowmem = 0xe0000000). This is the
- * default value for max_ram_below_4g now.
- *
- * - Then, to gigabyte align the memory, we move the split to 3G
- * (lowmem = 0xc0000000). But only in case we have to split in
- * the first place, i.e. ram_size is larger than (traditional)
- * lowmem. And for new machine types (gigabyte_align = true)
- * only, for live migration compatibility reasons.
- *
- * - Next the max-ram-below-4g option was added, which allowed to
- * reduce lowmem to a smaller value, to allow a larger PCI I/O
- * window below 4G. qemu doesn't enforce gigabyte alignment here,
- * but prints a warning.
- *
- * - Finally max-ram-below-4g got updated to also allow raising lowmem,
- * so legacy non-PAE guests can get as much memory as possible in
- * the 32bit address space below 4G.
- *
- * - Note that Xen has its own ram setup code in xen_ram_init(),
- * called via xen_hvm_init_pc().
- *
- * Examples:
- * qemu -M pc-1.7 -m 4G (old default) -> 3584M low, 512M high
- * qemu -M pc -m 4G (new default) -> 3072M low, 1024M high
- * qemu -M pc,max-ram-below-4g=2G -m 4G -> 2048M low, 2048M high
- * qemu -M pc,max-ram-below-4g=4G -m 3968M -> 3968M low (=4G-128M)
+ * There is no RAM split for the isapc machine
*/
if (xen_enabled()) {
xen_hvm_init_pc(pcms, &ram_memory);
} else {
ram_memory = machine->ram;
- if (!pcms->max_ram_below_4g) {
- pcms->max_ram_below_4g = 0xe0000000; /* default: 3.5G */
- }
- lowmem = pcms->max_ram_below_4g;
- if (machine->ram_size >= pcms->max_ram_below_4g) {
- if (pcmc->gigabyte_align) {
- if (lowmem > 0xc0000000) {
- lowmem = 0xc0000000;
- }
- if (lowmem & (1 * GiB - 1)) {
- warn_report("Large machine and max_ram_below_4g "
- "(%" PRIu64 ") not a multiple of 1G; "
- "possible bad performance.",
- pcms->max_ram_below_4g);
- }
- }
- }
- if (machine->ram_size >= lowmem) {
- x86ms->above_4g_mem_size = machine->ram_size - lowmem;
- x86ms->below_4g_mem_size = lowmem;
- } else {
- x86ms->above_4g_mem_size = 0;
- x86ms->below_4g_mem_size = machine->ram_size;
- }
+ pcms->max_ram_below_4g = 4 * GiB;
+ x86ms->above_4g_mem_size = 0;
+ x86ms->below_4g_mem_size = machine->ram_size;
}
/*
--
2.43.0On 8/22/2025 8:11 PM, Mark Cave-Ayland wrote:
> All isapc machines must have 32-bit CPUs and so the RAM split logic can be hardcoded
> accordingly.
>
> Signed-off-by: Mark Cave-Ayland <mark.caveayland@nutanix.com>
> ---
> hw/i386/pc_piix.c | 58 ++++-------------------------------------------
> 1 file changed, 4 insertions(+), 54 deletions(-)
>
> diff --git a/hw/i386/pc_piix.c b/hw/i386/pc_piix.c
> index 816124c027..fc94937ad4 100644
> --- a/hw/i386/pc_piix.c
> +++ b/hw/i386/pc_piix.c
> @@ -444,69 +444,19 @@ static void pc_init_isa(MachineState *machine)
> GSIState *gsi_state;
> MemoryRegion *ram_memory;
> MemoryRegion *rom_memory = system_memory;
> - ram_addr_t lowmem;
> uint64_t hole64_size = 0;
>
> /*
> - * Calculate ram split, for memory below and above 4G. It's a bit
> - * complicated for backward compatibility reasons ...
> - *
> - * - Traditional split is 3.5G (lowmem = 0xe0000000). This is the
> - * default value for max_ram_below_4g now.
> - *
> - * - Then, to gigabyte align the memory, we move the split to 3G
> - * (lowmem = 0xc0000000). But only in case we have to split in
> - * the first place, i.e. ram_size is larger than (traditional)
> - * lowmem. And for new machine types (gigabyte_align = true)
> - * only, for live migration compatibility reasons.
> - *
> - * - Next the max-ram-below-4g option was added, which allowed to
> - * reduce lowmem to a smaller value, to allow a larger PCI I/O
> - * window below 4G. qemu doesn't enforce gigabyte alignment here,
> - * but prints a warning.
> - *
> - * - Finally max-ram-below-4g got updated to also allow raising lowmem,
> - * so legacy non-PAE guests can get as much memory as possible in
> - * the 32bit address space below 4G.
> - *
> - * - Note that Xen has its own ram setup code in xen_ram_init(),
> - * called via xen_hvm_init_pc().
> - *
> - * Examples:
> - * qemu -M pc-1.7 -m 4G (old default) -> 3584M low, 512M high
> - * qemu -M pc -m 4G (new default) -> 3072M low, 1024M high
> - * qemu -M pc,max-ram-below-4g=2G -m 4G -> 2048M low, 2048M high
> - * qemu -M pc,max-ram-below-4g=4G -m 3968M -> 3968M low (=4G-128M)
> + * There is no RAM split for the isapc machine
> */
> if (xen_enabled()) {
> xen_hvm_init_pc(pcms, &ram_memory);
> } else {
> ram_memory = machine->ram;
> - if (!pcms->max_ram_below_4g) {
> - pcms->max_ram_below_4g = 0xe0000000; /* default: 3.5G */
> - }
> - lowmem = pcms->max_ram_below_4g;
> - if (machine->ram_size >= pcms->max_ram_below_4g) {
> - if (pcmc->gigabyte_align) {
> - if (lowmem > 0xc0000000) {
> - lowmem = 0xc0000000;
> - }
> - if (lowmem & (1 * GiB - 1)) {
> - warn_report("Large machine and max_ram_below_4g "
> - "(%" PRIu64 ") not a multiple of 1G; "
> - "possible bad performance.",
> - pcms->max_ram_below_4g);
> - }
> - }
> - }
>
> - if (machine->ram_size >= lowmem) {
> - x86ms->above_4g_mem_size = machine->ram_size - lowmem;
> - x86ms->below_4g_mem_size = lowmem;
> - } else {
> - x86ms->above_4g_mem_size = 0;
> - x86ms->below_4g_mem_size = machine->ram_size;
> - }
> + pcms->max_ram_below_4g = 4 * GiB;
> + x86ms->above_4g_mem_size = 0;
> + x86ms->below_4g_mem_size = machine->ram_size;
I think we need to sanity check the machine->ram_size is not bigger than
4G, and error out if it exceeds.
> }
>
> /*
On 26/08/2025 11:01, Xiaoyao Li wrote:
> On 8/22/2025 8:11 PM, Mark Cave-Ayland wrote:
>> All isapc machines must have 32-bit CPUs and so the RAM split logic
>> can be hardcoded
>> accordingly.
>>
>> Signed-off-by: Mark Cave-Ayland <mark.caveayland@nutanix.com>
>> ---
>> hw/i386/pc_piix.c | 58 ++++-------------------------------------------
>> 1 file changed, 4 insertions(+), 54 deletions(-)
>>
>> diff --git a/hw/i386/pc_piix.c b/hw/i386/pc_piix.c
>> index 816124c027..fc94937ad4 100644
>> --- a/hw/i386/pc_piix.c
>> +++ b/hw/i386/pc_piix.c
>> @@ -444,69 +444,19 @@ static void pc_init_isa(MachineState *machine)
>> GSIState *gsi_state;
>> MemoryRegion *ram_memory;
>> MemoryRegion *rom_memory = system_memory;
>> - ram_addr_t lowmem;
>> uint64_t hole64_size = 0;
>> /*
>> - * Calculate ram split, for memory below and above 4G. It's a bit
>> - * complicated for backward compatibility reasons ...
>> - *
>> - * - Traditional split is 3.5G (lowmem = 0xe0000000). This is the
>> - * default value for max_ram_below_4g now.
>> - *
>> - * - Then, to gigabyte align the memory, we move the split to 3G
>> - * (lowmem = 0xc0000000). But only in case we have to split in
>> - * the first place, i.e. ram_size is larger than (traditional)
>> - * lowmem. And for new machine types (gigabyte_align = true)
>> - * only, for live migration compatibility reasons.
>> - *
>> - * - Next the max-ram-below-4g option was added, which allowed to
>> - * reduce lowmem to a smaller value, to allow a larger PCI I/O
>> - * window below 4G. qemu doesn't enforce gigabyte alignment
>> here,
>> - * but prints a warning.
>> - *
>> - * - Finally max-ram-below-4g got updated to also allow raising
>> lowmem,
>> - * so legacy non-PAE guests can get as much memory as possible in
>> - * the 32bit address space below 4G.
>> - *
>> - * - Note that Xen has its own ram setup code in xen_ram_init(),
>> - * called via xen_hvm_init_pc().
>> - *
>> - * Examples:
>> - * qemu -M pc-1.7 -m 4G (old default) -> 3584M low,
>> 512M high
>> - * qemu -M pc -m 4G (new default) -> 3072M low,
>> 1024M high
>> - * qemu -M pc,max-ram-below-4g=2G -m 4G -> 2048M low,
>> 2048M high
>> - * qemu -M pc,max-ram-below-4g=4G -m 3968M -> 3968M low
>> (=4G-128M)
>> + * There is no RAM split for the isapc machine
>> */
>> if (xen_enabled()) {
>> xen_hvm_init_pc(pcms, &ram_memory);
>> } else {
>> ram_memory = machine->ram;
>> - if (!pcms->max_ram_below_4g) {
>> - pcms->max_ram_below_4g = 0xe0000000; /* default: 3.5G */
>> - }
>> - lowmem = pcms->max_ram_below_4g;
>> - if (machine->ram_size >= pcms->max_ram_below_4g) {
>> - if (pcmc->gigabyte_align) {
>> - if (lowmem > 0xc0000000) {
>> - lowmem = 0xc0000000;
>> - }
>> - if (lowmem & (1 * GiB - 1)) {
>> - warn_report("Large machine and max_ram_below_4g "
>> - "(%" PRIu64 ") not a multiple of 1G; "
>> - "possible bad performance.",
>> - pcms->max_ram_below_4g);
>> - }
>> - }
>> - }
>> - if (machine->ram_size >= lowmem) {
>> - x86ms->above_4g_mem_size = machine->ram_size - lowmem;
>> - x86ms->below_4g_mem_size = lowmem;
>> - } else {
>> - x86ms->above_4g_mem_size = 0;
>> - x86ms->below_4g_mem_size = machine->ram_size;
>> - }
>> + pcms->max_ram_below_4g = 4 * GiB;
>> + x86ms->above_4g_mem_size = 0;
>> + x86ms->below_4g_mem_size = machine->ram_size;
>
> I think we need to sanity check the machine->ram_size is not bigger than
> 4G, and error out if it exceeds.
Amazingly there is currently no limit for the isapc machine, but I shall
add it in for v7.
>> }
>> /*
ATB,
Mark.
On 27/08/2025 12:00, Mark Cave-Ayland wrote:
> On 26/08/2025 11:01, Xiaoyao Li wrote:
>
>> On 8/22/2025 8:11 PM, Mark Cave-Ayland wrote:
>>> All isapc machines must have 32-bit CPUs and so the RAM split logic
>>> can be hardcoded
>>> accordingly.
>>>
>>> Signed-off-by: Mark Cave-Ayland <mark.caveayland@nutanix.com>
>>> ---
>>> hw/i386/pc_piix.c | 58 ++++-------------------------------------------
>>> 1 file changed, 4 insertions(+), 54 deletions(-)
>>>
>>> diff --git a/hw/i386/pc_piix.c b/hw/i386/pc_piix.c
>>> index 816124c027..fc94937ad4 100644
>>> --- a/hw/i386/pc_piix.c
>>> +++ b/hw/i386/pc_piix.c
>>> @@ -444,69 +444,19 @@ static void pc_init_isa(MachineState *machine)
>>> GSIState *gsi_state;
>>> MemoryRegion *ram_memory;
>>> MemoryRegion *rom_memory = system_memory;
>>> - ram_addr_t lowmem;
>>> uint64_t hole64_size = 0;
>>> /*
>>> - * Calculate ram split, for memory below and above 4G. It's a bit
>>> - * complicated for backward compatibility reasons ...
>>> - *
>>> - * - Traditional split is 3.5G (lowmem = 0xe0000000). This is the
>>> - * default value for max_ram_below_4g now.
>>> - *
>>> - * - Then, to gigabyte align the memory, we move the split to 3G
>>> - * (lowmem = 0xc0000000). But only in case we have to split in
>>> - * the first place, i.e. ram_size is larger than (traditional)
>>> - * lowmem. And for new machine types (gigabyte_align = true)
>>> - * only, for live migration compatibility reasons.
>>> - *
>>> - * - Next the max-ram-below-4g option was added, which allowed to
>>> - * reduce lowmem to a smaller value, to allow a larger PCI I/O
>>> - * window below 4G. qemu doesn't enforce gigabyte alignment
>>> here,
>>> - * but prints a warning.
>>> - *
>>> - * - Finally max-ram-below-4g got updated to also allow raising
>>> lowmem,
>>> - * so legacy non-PAE guests can get as much memory as
>>> possible in
>>> - * the 32bit address space below 4G.
>>> - *
>>> - * - Note that Xen has its own ram setup code in xen_ram_init(),
>>> - * called via xen_hvm_init_pc().
>>> - *
>>> - * Examples:
>>> - * qemu -M pc-1.7 -m 4G (old default) -> 3584M low,
>>> 512M high
>>> - * qemu -M pc -m 4G (new default) -> 3072M low,
>>> 1024M high
>>> - * qemu -M pc,max-ram-below-4g=2G -m 4G -> 2048M low,
>>> 2048M high
>>> - * qemu -M pc,max-ram-below-4g=4G -m 3968M -> 3968M low
>>> (=4G-128M)
>>> + * There is no RAM split for the isapc machine
>>> */
>>> if (xen_enabled()) {
>>> xen_hvm_init_pc(pcms, &ram_memory);
>>> } else {
>>> ram_memory = machine->ram;
>>> - if (!pcms->max_ram_below_4g) {
>>> - pcms->max_ram_below_4g = 0xe0000000; /* default: 3.5G */
>>> - }
>>> - lowmem = pcms->max_ram_below_4g;
>>> - if (machine->ram_size >= pcms->max_ram_below_4g) {
>>> - if (pcmc->gigabyte_align) {
>>> - if (lowmem > 0xc0000000) {
>>> - lowmem = 0xc0000000;
>>> - }
>>> - if (lowmem & (1 * GiB - 1)) {
>>> - warn_report("Large machine and max_ram_below_4g "
>>> - "(%" PRIu64 ") not a multiple of 1G; "
>>> - "possible bad performance.",
>>> - pcms->max_ram_below_4g);
>>> - }
>>> - }
>>> - }
>>> - if (machine->ram_size >= lowmem) {
>>> - x86ms->above_4g_mem_size = machine->ram_size - lowmem;
>>> - x86ms->below_4g_mem_size = lowmem;
>>> - } else {
>>> - x86ms->above_4g_mem_size = 0;
>>> - x86ms->below_4g_mem_size = machine->ram_size;
>>> - }
>>> + pcms->max_ram_below_4g = 4 * GiB;
>>> + x86ms->above_4g_mem_size = 0;
>>> + x86ms->below_4g_mem_size = machine->ram_size;
>>
>> I think we need to sanity check the machine->ram_size is not bigger
>> than 4G, and error out if it exceeds.
>
> Amazingly there is currently no limit for the isapc machine, but I shall
> add it in for v7.
With the PCI hole removed it appears that TCG and KVM have a different
idea as to the maximum allowable amount of RAM available:
Fails with KVM:
./build/qemu-system-x86_64 -accel kvm -M isapc -m 4079M
qemu-system-x86_64: kvm_set_user_memory_region:
KVM_SET_USER_MEMORY_REGION failed, slot=0, start=0x0, size=0xfef00000:
File exists
kvm_set_phys_mem: error registering slot: File exists
Works with TCG:
./build/qemu-system-x86_64 -accel tcg -M isapc -m 4096M
Given that the original limit above is 3.5G I think it's best to revert
back to using 3.5G instead of 4G so that both TCG and KVM behave the
same, whilst also allowing a bit of wiggle room if required.
ATB,
Mark.
On 28/8/25 10:41, Mark Cave-Ayland wrote:
> On 27/08/2025 12:00, Mark Cave-Ayland wrote:
>
>> On 26/08/2025 11:01, Xiaoyao Li wrote:
>>
>>> On 8/22/2025 8:11 PM, Mark Cave-Ayland wrote:
>>>> All isapc machines must have 32-bit CPUs and so the RAM split logic
>>>> can be hardcoded
>>>> accordingly.
>>>>
>>>> Signed-off-by: Mark Cave-Ayland <mark.caveayland@nutanix.com>
>>>> ---
>>>> hw/i386/pc_piix.c | 58 +++
>>>> +-------------------------------------------
>>>> 1 file changed, 4 insertions(+), 54 deletions(-)
>>>>
>>>> diff --git a/hw/i386/pc_piix.c b/hw/i386/pc_piix.c
>>>> index 816124c027..fc94937ad4 100644
>>>> --- a/hw/i386/pc_piix.c
>>>> +++ b/hw/i386/pc_piix.c
>>>> @@ -444,69 +444,19 @@ static void pc_init_isa(MachineState *machine)
>>>> GSIState *gsi_state;
>>>> MemoryRegion *ram_memory;
>>>> MemoryRegion *rom_memory = system_memory;
>>>> - ram_addr_t lowmem;
>>>> uint64_t hole64_size = 0;
>>>> /*
>>>> - * Calculate ram split, for memory below and above 4G. It's a bit
>>>> - * complicated for backward compatibility reasons ...
>>>> - *
>>>> - * - Traditional split is 3.5G (lowmem = 0xe0000000). This is
>>>> the
>>>> - * default value for max_ram_below_4g now.
>>>> - *
>>>> - * - Then, to gigabyte align the memory, we move the split to 3G
>>>> - * (lowmem = 0xc0000000). But only in case we have to split in
>>>> - * the first place, i.e. ram_size is larger than (traditional)
>>>> - * lowmem. And for new machine types (gigabyte_align = true)
>>>> - * only, for live migration compatibility reasons.
>>>> - *
>>>> - * - Next the max-ram-below-4g option was added, which allowed to
>>>> - * reduce lowmem to a smaller value, to allow a larger PCI I/O
>>>> - * window below 4G. qemu doesn't enforce gigabyte alignment
>>>> here,
>>>> - * but prints a warning.
>>>> - *
>>>> - * - Finally max-ram-below-4g got updated to also allow
>>>> raising lowmem,
>>>> - * so legacy non-PAE guests can get as much memory as
>>>> possible in
>>>> - * the 32bit address space below 4G.
>>>> - *
>>>> - * - Note that Xen has its own ram setup code in xen_ram_init(),
>>>> - * called via xen_hvm_init_pc().
>>>> - *
>>>> - * Examples:
>>>> - * qemu -M pc-1.7 -m 4G (old default) -> 3584M low,
>>>> 512M high
>>>> - * qemu -M pc -m 4G (new default) -> 3072M low,
>>>> 1024M high
>>>> - * qemu -M pc,max-ram-below-4g=2G -m 4G -> 2048M low,
>>>> 2048M high
>>>> - * qemu -M pc,max-ram-below-4g=4G -m 3968M -> 3968M low
>>>> (=4G-128M)
>>>> + * There is no RAM split for the isapc machine
>>>> */
>>>> if (xen_enabled()) {
>>>> xen_hvm_init_pc(pcms, &ram_memory);
>>>> } else {
>>>> ram_memory = machine->ram;
>>>> - if (!pcms->max_ram_below_4g) {
>>>> - pcms->max_ram_below_4g = 0xe0000000; /* default: 3.5G */
>>>> - }
>>>> - lowmem = pcms->max_ram_below_4g;
>>>> - if (machine->ram_size >= pcms->max_ram_below_4g) {
>>>> - if (pcmc->gigabyte_align) {
>>>> - if (lowmem > 0xc0000000) {
>>>> - lowmem = 0xc0000000;
>>>> - }
>>>> - if (lowmem & (1 * GiB - 1)) {
>>>> - warn_report("Large machine and max_ram_below_4g "
>>>> - "(%" PRIu64 ") not a multiple of 1G; "
>>>> - "possible bad performance.",
>>>> - pcms->max_ram_below_4g);
>>>> - }
>>>> - }
>>>> - }
>>>> - if (machine->ram_size >= lowmem) {
>>>> - x86ms->above_4g_mem_size = machine->ram_size - lowmem;
>>>> - x86ms->below_4g_mem_size = lowmem;
>>>> - } else {
>>>> - x86ms->above_4g_mem_size = 0;
>>>> - x86ms->below_4g_mem_size = machine->ram_size;
>>>> - }
>>>> + pcms->max_ram_below_4g = 4 * GiB;
>>>> + x86ms->above_4g_mem_size = 0;
>>>> + x86ms->below_4g_mem_size = machine->ram_size;
>>>
>>> I think we need to sanity check the machine->ram_size is not bigger
>>> than 4G, and error out if it exceeds.
>>
>> Amazingly there is currently no limit for the isapc machine, but I
>> shall add it in for v7.
>
> With the PCI hole removed it appears that TCG and KVM have a different
> idea as to the maximum allowable amount of RAM available:
>
> Fails with KVM:
> ./build/qemu-system-x86_64 -accel kvm -M isapc -m 4079M
> qemu-system-x86_64: kvm_set_user_memory_region:
> KVM_SET_USER_MEMORY_REGION failed, slot=0, start=0x0, size=0xfef00000:
> File exists
> kvm_set_phys_mem: error registering slot: File exists
Is this a bug? Clearly not obvious error message...
>
> Works with TCG:
> ./build/qemu-system-x86_64 -accel tcg -M isapc -m 4096M
>
> Given that the original limit above is 3.5G I think it's best to revert
> back to using 3.5G instead of 4G so that both TCG and KVM behave the
> same, whilst also allowing a bit of wiggle room if required.
>
>
> ATB,
>
> Mark.
>
>
On 22/09/2025 13:08, Philippe Mathieu-Daudé wrote:
> On 28/8/25 10:41, Mark Cave-Ayland wrote:
>> On 27/08/2025 12:00, Mark Cave-Ayland wrote:
>>
>>> On 26/08/2025 11:01, Xiaoyao Li wrote:
>>>
>>>> On 8/22/2025 8:11 PM, Mark Cave-Ayland wrote:
>>>>> All isapc machines must have 32-bit CPUs and so the RAM split logic
>>>>> can be hardcoded
>>>>> accordingly.
>>>>>
>>>>> Signed-off-by: Mark Cave-Ayland <mark.caveayland@nutanix.com>
>>>>> ---
>>>>> hw/i386/pc_piix.c | 58 +++
>>>>> +-------------------------------------------
>>>>> 1 file changed, 4 insertions(+), 54 deletions(-)
>>>>>
>>>>> diff --git a/hw/i386/pc_piix.c b/hw/i386/pc_piix.c
>>>>> index 816124c027..fc94937ad4 100644
>>>>> --- a/hw/i386/pc_piix.c
>>>>> +++ b/hw/i386/pc_piix.c
>>>>> @@ -444,69 +444,19 @@ static void pc_init_isa(MachineState *machine)
>>>>> GSIState *gsi_state;
>>>>> MemoryRegion *ram_memory;
>>>>> MemoryRegion *rom_memory = system_memory;
>>>>> - ram_addr_t lowmem;
>>>>> uint64_t hole64_size = 0;
>>>>> /*
>>>>> - * Calculate ram split, for memory below and above 4G. It's a
>>>>> bit
>>>>> - * complicated for backward compatibility reasons ...
>>>>> - *
>>>>> - * - Traditional split is 3.5G (lowmem = 0xe0000000). This
>>>>> is the
>>>>> - * default value for max_ram_below_4g now.
>>>>> - *
>>>>> - * - Then, to gigabyte align the memory, we move the split to 3G
>>>>> - * (lowmem = 0xc0000000). But only in case we have to
>>>>> split in
>>>>> - * the first place, i.e. ram_size is larger than (traditional)
>>>>> - * lowmem. And for new machine types (gigabyte_align = true)
>>>>> - * only, for live migration compatibility reasons.
>>>>> - *
>>>>> - * - Next the max-ram-below-4g option was added, which
>>>>> allowed to
>>>>> - * reduce lowmem to a smaller value, to allow a larger PCI I/O
>>>>> - * window below 4G. qemu doesn't enforce gigabyte
>>>>> alignment here,
>>>>> - * but prints a warning.
>>>>> - *
>>>>> - * - Finally max-ram-below-4g got updated to also allow
>>>>> raising lowmem,
>>>>> - * so legacy non-PAE guests can get as much memory as
>>>>> possible in
>>>>> - * the 32bit address space below 4G.
>>>>> - *
>>>>> - * - Note that Xen has its own ram setup code in xen_ram_init(),
>>>>> - * called via xen_hvm_init_pc().
>>>>> - *
>>>>> - * Examples:
>>>>> - * qemu -M pc-1.7 -m 4G (old default) -> 3584M low,
>>>>> 512M high
>>>>> - * qemu -M pc -m 4G (new default) -> 3072M low,
>>>>> 1024M high
>>>>> - * qemu -M pc,max-ram-below-4g=2G -m 4G -> 2048M low,
>>>>> 2048M high
>>>>> - * qemu -M pc,max-ram-below-4g=4G -m 3968M -> 3968M low
>>>>> (=4G-128M)
>>>>> + * There is no RAM split for the isapc machine
>>>>> */
>>>>> if (xen_enabled()) {
>>>>> xen_hvm_init_pc(pcms, &ram_memory);
>>>>> } else {
>>>>> ram_memory = machine->ram;
>>>>> - if (!pcms->max_ram_below_4g) {
>>>>> - pcms->max_ram_below_4g = 0xe0000000; /* default: 3.5G */
>>>>> - }
>>>>> - lowmem = pcms->max_ram_below_4g;
>>>>> - if (machine->ram_size >= pcms->max_ram_below_4g) {
>>>>> - if (pcmc->gigabyte_align) {
>>>>> - if (lowmem > 0xc0000000) {
>>>>> - lowmem = 0xc0000000;
>>>>> - }
>>>>> - if (lowmem & (1 * GiB - 1)) {
>>>>> - warn_report("Large machine and max_ram_below_4g "
>>>>> - "(%" PRIu64 ") not a multiple of
>>>>> 1G; "
>>>>> - "possible bad performance.",
>>>>> - pcms->max_ram_below_4g);
>>>>> - }
>>>>> - }
>>>>> - }
>>>>> - if (machine->ram_size >= lowmem) {
>>>>> - x86ms->above_4g_mem_size = machine->ram_size - lowmem;
>>>>> - x86ms->below_4g_mem_size = lowmem;
>>>>> - } else {
>>>>> - x86ms->above_4g_mem_size = 0;
>>>>> - x86ms->below_4g_mem_size = machine->ram_size;
>>>>> - }
>>>>> + pcms->max_ram_below_4g = 4 * GiB;
>>>>> + x86ms->above_4g_mem_size = 0;
>>>>> + x86ms->below_4g_mem_size = machine->ram_size;
>>>>
>>>> I think we need to sanity check the machine->ram_size is not bigger
>>>> than 4G, and error out if it exceeds.
>>>
>>> Amazingly there is currently no limit for the isapc machine, but I
>>> shall add it in for v7.
>>
>> With the PCI hole removed it appears that TCG and KVM have a different
>> idea as to the maximum allowable amount of RAM available:
>>
>> Fails with KVM:
>> ./build/qemu-system-x86_64 -accel kvm -M isapc -m 4079M
>> qemu-system-x86_64: kvm_set_user_memory_region:
>> KVM_SET_USER_MEMORY_REGION failed, slot=0, start=0x0, size=0xfef00000:
>> File exists
>> kvm_set_phys_mem: error registering slot: File exists
>
> Is this a bug? Clearly not obvious error message...
I'm not sure? I suspect the change to the PCI hole size is what causes
this to appear, but I'm not overly familiar with the KVM memory layout.
ATB,
Mark.
On 22/8/25 14:11, Mark Cave-Ayland wrote: > All isapc machines must have 32-bit CPUs and so the RAM split logic can be hardcoded > accordingly. > > Signed-off-by: Mark Cave-Ayland <mark.caveayland@nutanix.com> > --- > hw/i386/pc_piix.c | 58 ++++------------------------------------------- > 1 file changed, 4 insertions(+), 54 deletions(-) Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
© 2016 - 2025 Red Hat, Inc.