Scroll to navigation

PMEM_FLUSH(3) PMDK Programmer's Manual PMEM_FLUSH(3)

NAME

pmem_flush(), pmem_drain(), pmem_persist(), pmem_msync(), pmem_deep_flush(), pmem_deep_drain(), pmem_deep_persist(), pmem_has_hw_drain(), pmem_has_auto_flush() - check persistency, store persistent data and delete mappings

SYNOPSIS


#include <libpmem.h>
void pmem_persist(const void *addr, size_t len);
int pmem_msync(const void *addr, size_t len);
void pmem_flush(const void *addr, size_t len);
void pmem_deep_flush(const void *addr, size_t len); (EXPERIMENTAL)
int pmem_deep_drain(const void *addr, size_t len); (EXPERIMENTAL)
int pmem_deep_persist(const void *addr, size_t len); (EXPERIMENTAL)
void pmem_drain(void);
int pmem_has_auto_flush(void); (EXPERIMENTAL)
int pmem_has_hw_drain(void);

    

DESCRIPTION

The functions in this section provide access to the stages of flushing to persistence, for the less common cases where an application needs more control of the flushing operations than the pmem_persist() function.

WARNING: Using pmem_persist() on a range where pmem_is_pmem(3) returns false may not do anything useful – use msync(2) instead.

The pmem_persist() function force any changes in the range [addr, addr+len) to be stored durably in persistent memory. This is equivalent to calling msync(2) but may be more optimal and will avoid calling into the kernel if possible. There are no alignment restrictions on the range described by addr and len, but pmem_persist() may expand the range as necessary to meet platform alignment requirements.

WARNING: Like msync(2), there is nothing atomic or transactional about this call. Any unwritten stores in the given range will be written, but some stores may have already been written by virtue of normal cache eviction/replacement policies. Correctly written code must not depend on stores waiting until pmem_persist() is called to become persistent – they can become persistent at any time before pmem_persist() is called.

The pmem_msync() function is like pmem_persist() in that it forces any changes in the range [addr, addr+len) to be stored durably. Since it calls msync(), this function works on either persistent memory or a memory mapped file on traditional storage. pmem_msync() takes steps to ensure the alignment of addresses and lengths passed to msync() meet the requirements of that system call. It calls msync() with the MS_SYNC flag as described in msync(2). Typically the application only checks for the existence of persistent memory once, and then uses that result throughout the program, for example:


/* do this call once, after the pmem is memory mapped */
int is_pmem = pmem_is_pmem(rangeaddr, rangelen);
/* ... make changes to a range of pmem ... */
/* make the changes durable */
if (is_pmem)
    pmem_persist(subrangeaddr, subrangelen);
else
    pmem_msync(subrangeaddr, subrangelen);
/* ... */

    

WARNING: On Linux, pmem_msync() and msync(2) have no effect on memory ranges mapped from Device DAX. In case of memory ranges where pmem_is_pmem(3) returns true use pmem_persist() to force the changes to be stored durably in persistent memory.

The pmem_flush() and pmem_drain() functions provide partial versions of the pmem_persist() function. pmem_persist() can be thought of as this:


void
pmem_persist(const void *addr, size_t len)
{
    /* flush the processor caches */
    pmem_flush(addr, len);
    /* wait for any pmem stores to drain from HW buffers */
    pmem_drain();
}

    

These functions allow advanced programs to create their own variations of pmem_persist(). For example, a program that needs to flush several discontiguous ranges can call pmem_flush() for each range and then follow up by calling pmem_drain() once.

The semantics of pmem_deep_flush() function is the same as pmem_flush() function except that pmem_deep_flush() is indifferent to PMEM_NO_FLUSH environment variable (see ENVIRONMENT section in libpmem(7)) and always flushes processor caches.

The behavior of pmem_deep_persist() function is the same as pmem_persist(), except that it provides higher reliability by flushing persistent memory stores to the most reliable persistence domain available to software rather than depending on automatic WPQ (write pending queue) flush on power failure (ADR).

The pmem_deep_flush() and pmem_deep_drain() functions provide partial versions of pmem_deep_persist() function. pmem_deep_persist() can be thought of as this:


int pmem_deep_persist(const void *addr, size_t len)
{
    /* flush the processor caches */
    pmem_deep_flush(addr, len);
    /* wait for any pmem stores to drain from HW buffers */
    return pmem_deep_drain(addr, len);
}

    

Since this operation is usually much more expensive than pmem_persist(), it should be used rarely. Typically the application should use this function only to flush the most critical data, which are required to recover after the power failure.

The pmem_has_auto_flush() function checks if the machine supports automatic CPU cache flush on power failure or system crash. Function returns true only when each NVDIMM in the system is covered by this mechanism.

The pmem_has_hw_drain() function checks if the machine supports an explicit hardware drain instruction for persistent memory.

RETURN VALUE

The pmem_persist() function returns no value.

The pmem_msync() return value is the return value of msync(), which can return -1 and set errno to indicate an error.

The pmem_flush(), pmem_drain() and pmem_deep_flush() functions return no value.

The pmem_deep_persist() and pmem_deep_drain() return 0 on success. Otherwise it returns -1 and sets errno appropriately. If len is equal zero pmem_deep_persist() and pmem_deep_drain() return 0 but no flushing take place.

The pmem_has_auto_flush() function returns 1 if given platform supports processor cache flushing on a power loss event. Otherwise it returns 0. On error it returns -1 and sets errno appropriately.

The pmem_has_hw_drain() function returns true if the machine supports an explicit hardware drain instruction for persistent memory. On Intel processors with persistent memory, stores to persistent memory are considered persistent once they are flushed from the CPU caches, so this function always returns false. Despite that, programs using pmem_flush() to flush ranges of memory should still follow up by calling pmem_drain() once to ensure the flushes are complete. As mentioned above, pmem_persist() handles calling both pmem_flush() and pmem_drain().

SEE ALSO

msync(2), pmem_is_pmem(3), libpmem(7) and <https://pmem.io>
2020-09-16 PMDK - pmem API version 1.1