Re: [Qemu-devel] [PATCH qemu 1/2] exec, kvm, target-ppc: Move getrampagesize() to common code

[Alexey Kardashevskiy]

On 10/02/17 11:41, David Gibson wrote:
> On Thu, Feb 09, 2017 at 12:48:19PM +0100, Paolo Bonzini wrote:
>>
>>
>> On 09/02/2017 06:43, Alexey Kardashevskiy wrote:
>>> On 03/01/17 10:34, David Gibson wrote:
>>>> On Thu, Dec 22, 2016 at 04:22:11PM +1100, Alexey Kardashevskiy wrote:
>>>>> getrampagesize() returns the largest supported page size and mainly
>>>>> used to know if huge pages are enabled.
>>>>>
>>>>> However is implemented in target-ppc/kvm.c and not available
>>>>> in TCG or other architectures.
>>>>>
>>>>> This renames and moves gethugepagesize() to mmap-alloc.c where
>>>>> fd-based analog of it is already implemented. This renames and moves
>>>>> getrampagesize() to exec.c as it seems to be the common place for
>>>>> helpers like this.
>>>>>
>>>>> This first user for it is going to be a spapr-pci-host-bridge which
>>>>> needs to know the largest RAM page size so the guest could try
>>>>> using bigger IOMMU pages to save memory.
>>>>>
>>>>> Signed-off-by: Alexey Kardashevskiy <address@hidden>
>>>>
>>>> Reviewed-by: David Gibson <address@hidden>
>>>>
>>>> Seems sensible to me, but I'm not comfortable merging this via my tree
>>>> since it touches such core code.  Probably should go via Paolo.
>>>
>>> Paolo, ping?
>>
>> It's just code movement, go ahead.
> 
> Ok, I've merged this in my tree.


I cannot find it in any public tree, where did it go?



> 
> 
>>
>> Paolo
>>
>>>
>>>
>>>>
>>>>> ---
>>>>>  include/exec/ram_addr.h   |   1 +
>>>>>  include/qemu/mmap-alloc.h |   2 +
>>>>>  exec.c                    |  82 ++++++++++++++++++++++++++++++++++++
>>>>>  target-ppc/kvm.c          | 105 
>>>>> ++--------------------------------------------
>>>>>  util/mmap-alloc.c         |  25 +++++++++++
>>>>>  5 files changed, 113 insertions(+), 102 deletions(-)
>>>>>
>>>>> diff --git a/include/exec/ram_addr.h b/include/exec/ram_addr.h
>>>>> index 54d7108a9e..3935cbcfcd 100644
>>>>> --- a/include/exec/ram_addr.h
>>>>> +++ b/include/exec/ram_addr.h
>>>>> @@ -91,6 +91,7 @@ typedef struct RAMList {
>>>>>  } RAMList;
>>>>>  extern RAMList ram_list;
>>>>>  
>>>>> +long qemu_getrampagesize(void);
>>>>>  ram_addr_t last_ram_offset(void);
>>>>>  void qemu_mutex_lock_ramlist(void);
>>>>>  void qemu_mutex_unlock_ramlist(void);
>>>>> diff --git a/include/qemu/mmap-alloc.h b/include/qemu/mmap-alloc.h
>>>>> index 933c024ac5..50385e3f81 100644
>>>>> --- a/include/qemu/mmap-alloc.h
>>>>> +++ b/include/qemu/mmap-alloc.h
>>>>> @@ -5,6 +5,8 @@
>>>>>  
>>>>>  size_t qemu_fd_getpagesize(int fd);
>>>>>  
>>>>> +size_t qemu_mempath_getpagesize(const char *mem_path);
>>>>> +
>>>>>  void *qemu_ram_mmap(int fd, size_t size, size_t align, bool shared);
>>>>>  
>>>>>  void qemu_ram_munmap(void *ptr, size_t size);
>>>>> diff --git a/exec.c b/exec.c
>>>>> index 08c558eecf..d73b477a70 100644
>>>>> --- a/exec.c
>>>>> +++ b/exec.c
>>>>> @@ -32,6 +32,7 @@
>>>>>  #endif
>>>>>  #include "sysemu/kvm.h"
>>>>>  #include "sysemu/sysemu.h"
>>>>> +#include "sysemu/numa.h"
>>>>>  #include "qemu/timer.h"
>>>>>  #include "qemu/config-file.h"
>>>>>  #include "qemu/error-report.h"
>>>>> @@ -1218,6 +1219,87 @@ void qemu_mutex_unlock_ramlist(void)
>>>>>  }
>>>>>  
>>>>>  #ifdef __linux__
>>>>> +/*
>>>>> + * FIXME TOCTTOU: this iterates over memory backends' mem-path, which
>>>>> + * may or may not name the same files / on the same filesystem now as
>>>>> + * when we actually open and map them.  Iterate over the file
>>>>> + * descriptors instead, and use qemu_fd_getpagesize().
>>>>> + */
>>>>> +static int find_max_supported_pagesize(Object *obj, void *opaque)
>>>>> +{
>>>>> +    char *mem_path;
>>>>> +    long *hpsize_min = opaque;
>>>>> +
>>>>> +    if (object_dynamic_cast(obj, TYPE_MEMORY_BACKEND)) {
>>>>> +        mem_path = object_property_get_str(obj, "mem-path", NULL);
>>>>> +        if (mem_path) {
>>>>> +            long hpsize = qemu_mempath_getpagesize(mem_path);
>>>>> +            if (hpsize < *hpsize_min) {
>>>>> +                *hpsize_min = hpsize;
>>>>> +            }
>>>>> +        } else {
>>>>> +            *hpsize_min = getpagesize();
>>>>> +        }
>>>>> +    }
>>>>> +
>>>>> +    return 0;
>>>>> +}
>>>>> +
>>>>> +long qemu_getrampagesize(void)
>>>>> +{
>>>>> +    long hpsize = LONG_MAX;
>>>>> +    long mainrampagesize;
>>>>> +    Object *memdev_root;
>>>>> +
>>>>> +    if (mem_path) {
>>>>> +        mainrampagesize = qemu_mempath_getpagesize(mem_path);
>>>>> +    } else {
>>>>> +        mainrampagesize = getpagesize();
>>>>> +    }
>>>>> +
>>>>> +    /* it's possible we have memory-backend objects with
>>>>> +     * hugepage-backed RAM. these may get mapped into system
>>>>> +     * address space via -numa parameters or memory hotplug
>>>>> +     * hooks. we want to take these into account, but we
>>>>> +     * also want to make sure these supported hugepage
>>>>> +     * sizes are applicable across the entire range of memory
>>>>> +     * we may boot from, so we take the min across all
>>>>> +     * backends, and assume normal pages in cases where a
>>>>> +     * backend isn't backed by hugepages.
>>>>> +     */
>>>>> +    memdev_root = object_resolve_path("/objects", NULL);
>>>>> +    if (memdev_root) {
>>>>> +        object_child_foreach(memdev_root, find_max_supported_pagesize, 
>>>>> &hpsize);
>>>>> +    }
>>>>> +    if (hpsize == LONG_MAX) {
>>>>> +        /* No additional memory regions found ==> Report main RAM page 
>>>>> size */
>>>>> +        return mainrampagesize;
>>>>> +    }
>>>>> +
>>>>> +    /* If NUMA is disabled or the NUMA nodes are not backed with a
>>>>> +     * memory-backend, then there is at least one node using "normal" 
>>>>> RAM,
>>>>> +     * so if its page size is smaller we have got to report that size 
>>>>> instead.
>>>>> +     */
>>>>> +    if (hpsize > mainrampagesize &&
>>>>> +        (nb_numa_nodes == 0 || numa_info[0].node_memdev == NULL)) {
>>>>> +        static bool warned;
>>>>> +        if (!warned) {
>>>>> +            error_report("Huge page support disabled (n/a for main 
>>>>> memory).");
>>>>> +            warned = true;
>>>>> +        }
>>>>> +        return mainrampagesize;
>>>>> +    }
>>>>> +
>>>>> +    return hpsize;
>>>>> +}
>>>>> +#else
>>>>> +long qemu_getrampagesize(void)
>>>>> +{
>>>>> +    return getpagesize();
>>>>> +}
>>>>> +#endif
>>>>> +
>>>>> +#ifdef __linux__
>>>>>  static int64_t get_file_size(int fd)
>>>>>  {
>>>>>      int64_t size = lseek(fd, 0, SEEK_END);
>>>>> diff --git a/target-ppc/kvm.c b/target-ppc/kvm.c
>>>>> index 6e91a4d8bb..e0abffa8ad 100644
>>>>> --- a/target-ppc/kvm.c
>>>>> +++ b/target-ppc/kvm.c
>>>>> @@ -42,6 +42,7 @@
>>>>>  #include "trace.h"
>>>>>  #include "exec/gdbstub.h"
>>>>>  #include "exec/memattrs.h"
>>>>> +#include "exec/ram_addr.h"
>>>>>  #include "sysemu/hostmem.h"
>>>>>  #include "qemu/cutils.h"
>>>>>  #if defined(TARGET_PPC64)
>>>>> @@ -325,106 +326,6 @@ static void kvm_get_smmu_info(PowerPCCPU *cpu, 
>>>>> struct kvm_ppc_smmu_info *info)
>>>>>      kvm_get_fallback_smmu_info(cpu, info);
>>>>>  }
>>>>>  
>>>>> -static long gethugepagesize(const char *mem_path)
>>>>> -{
>>>>> -    struct statfs fs;
>>>>> -    int ret;
>>>>> -
>>>>> -    do {
>>>>> -        ret = statfs(mem_path, &fs);
>>>>> -    } while (ret != 0 && errno == EINTR);
>>>>> -
>>>>> -    if (ret != 0) {
>>>>> -        fprintf(stderr, "Couldn't statfs() memory path: %s\n",
>>>>> -                strerror(errno));
>>>>> -        exit(1);
>>>>> -    }
>>>>> -
>>>>> -#define HUGETLBFS_MAGIC       0x958458f6
>>>>> -
>>>>> -    if (fs.f_type != HUGETLBFS_MAGIC) {
>>>>> -        /* Explicit mempath, but it's ordinary pages */
>>>>> -        return getpagesize();
>>>>> -    }
>>>>> -
>>>>> -    /* It's hugepage, return the huge page size */
>>>>> -    return fs.f_bsize;
>>>>> -}
>>>>> -
>>>>> -/*
>>>>> - * FIXME TOCTTOU: this iterates over memory backends' mem-path, which
>>>>> - * may or may not name the same files / on the same filesystem now as
>>>>> - * when we actually open and map them.  Iterate over the file
>>>>> - * descriptors instead, and use qemu_fd_getpagesize().
>>>>> - */
>>>>> -static int find_max_supported_pagesize(Object *obj, void *opaque)
>>>>> -{
>>>>> -    char *mem_path;
>>>>> -    long *hpsize_min = opaque;
>>>>> -
>>>>> -    if (object_dynamic_cast(obj, TYPE_MEMORY_BACKEND)) {
>>>>> -        mem_path = object_property_get_str(obj, "mem-path", NULL);
>>>>> -        if (mem_path) {
>>>>> -            long hpsize = gethugepagesize(mem_path);
>>>>> -            if (hpsize < *hpsize_min) {
>>>>> -                *hpsize_min = hpsize;
>>>>> -            }
>>>>> -        } else {
>>>>> -            *hpsize_min = getpagesize();
>>>>> -        }
>>>>> -    }
>>>>> -
>>>>> -    return 0;
>>>>> -}
>>>>> -
>>>>> -static long getrampagesize(void)
>>>>> -{
>>>>> -    long hpsize = LONG_MAX;
>>>>> -    long mainrampagesize;
>>>>> -    Object *memdev_root;
>>>>> -
>>>>> -    if (mem_path) {
>>>>> -        mainrampagesize = gethugepagesize(mem_path);
>>>>> -    } else {
>>>>> -        mainrampagesize = getpagesize();
>>>>> -    }
>>>>> -
>>>>> -    /* it's possible we have memory-backend objects with
>>>>> -     * hugepage-backed RAM. these may get mapped into system
>>>>> -     * address space via -numa parameters or memory hotplug
>>>>> -     * hooks. we want to take these into account, but we
>>>>> -     * also want to make sure these supported hugepage
>>>>> -     * sizes are applicable across the entire range of memory
>>>>> -     * we may boot from, so we take the min across all
>>>>> -     * backends, and assume normal pages in cases where a
>>>>> -     * backend isn't backed by hugepages.
>>>>> -     */
>>>>> -    memdev_root = object_resolve_path("/objects", NULL);
>>>>> -    if (memdev_root) {
>>>>> -        object_child_foreach(memdev_root, find_max_supported_pagesize, 
>>>>> &hpsize);
>>>>> -    }
>>>>> -    if (hpsize == LONG_MAX) {
>>>>> -        /* No additional memory regions found ==> Report main RAM page 
>>>>> size */
>>>>> -        return mainrampagesize;
>>>>> -    }
>>>>> -
>>>>> -    /* If NUMA is disabled or the NUMA nodes are not backed with a
>>>>> -     * memory-backend, then there is at least one node using "normal" 
>>>>> RAM,
>>>>> -     * so if its page size is smaller we have got to report that size 
>>>>> instead.
>>>>> -     */
>>>>> -    if (hpsize > mainrampagesize &&
>>>>> -        (nb_numa_nodes == 0 || numa_info[0].node_memdev == NULL)) {
>>>>> -        static bool warned;
>>>>> -        if (!warned) {
>>>>> -            error_report("Huge page support disabled (n/a for main 
>>>>> memory).");
>>>>> -            warned = true;
>>>>> -        }
>>>>> -        return mainrampagesize;
>>>>> -    }
>>>>> -
>>>>> -    return hpsize;
>>>>> -}
>>>>> -
>>>>>  static bool kvm_valid_page_size(uint32_t flags, long rampgsize, uint32_t 
>>>>> shift)
>>>>>  {
>>>>>      if (!(flags & KVM_PPC_PAGE_SIZES_REAL)) {
>>>>> @@ -454,7 +355,7 @@ static void kvm_fixup_page_sizes(PowerPCCPU *cpu)
>>>>>          has_smmu_info = true;
>>>>>      }
>>>>>  
>>>>> -    rampagesize = getrampagesize();
>>>>> +    rampagesize = qemu_getrampagesize();
>>>>>  
>>>>>      /* Convert to QEMU form */
>>>>>      memset(&env->sps, 0, sizeof(env->sps));
>>>>> @@ -2177,7 +2078,7 @@ uint64_t kvmppc_rma_size(uint64_t current_size, 
>>>>> unsigned int hash_shift)
>>>>>      /* Find the largest hardware supported page size that's less than
>>>>>       * or equal to the (logical) backing page size of guest RAM */
>>>>>      kvm_get_smmu_info(POWERPC_CPU(first_cpu), &info);
>>>>> -    rampagesize = getrampagesize();
>>>>> +    rampagesize = qemu_getrampagesize();
>>>>>      best_page_shift = 0;
>>>>>  
>>>>>      for (i = 0; i < KVM_PPC_PAGE_SIZES_MAX_SZ; i++) {
>>>>> diff --git a/util/mmap-alloc.c b/util/mmap-alloc.c
>>>>> index 5a85aa3c89..564c79109c 100644
>>>>> --- a/util/mmap-alloc.c
>>>>> +++ b/util/mmap-alloc.c
>>>>> @@ -39,6 +39,31 @@ size_t qemu_fd_getpagesize(int fd)
>>>>>      return getpagesize();
>>>>>  }
>>>>>  
>>>>> +size_t qemu_mempath_getpagesize(const char *mem_path)
>>>>> +{
>>>>> +#ifdef CONFIG_LINUX
>>>>> +    struct statfs fs;
>>>>> +    int ret;
>>>>> +
>>>>> +    do {
>>>>> +        ret = statfs(mem_path, &fs);
>>>>> +    } while (ret != 0 && errno == EINTR);
>>>>> +
>>>>> +    if (ret != 0) {
>>>>> +        fprintf(stderr, "Couldn't statfs() memory path: %s\n",
>>>>> +                strerror(errno));
>>>>> +        exit(1);
>>>>> +    }
>>>>> +
>>>>> +    if (fs.f_type == HUGETLBFS_MAGIC) {
>>>>> +        /* It's hugepage, return the huge page size */
>>>>> +        return fs.f_bsize;
>>>>> +    }
>>>>> +#endif
>>>>> +
>>>>> +    return getpagesize();
>>>>> +}
>>>>> +
>>>>>  void *qemu_ram_mmap(int fd, size_t size, size_t align, bool shared)
>>>>>  {
>>>>>      /*
>>>>
>>>
>>>
>>
> 
> 
> 
> 


-- 
Alexey

signature.asc
Description: OpenPGP digital signature