d3dd927de3
This is based on some work in external benchmarks. What's worked well
there is emulating a global simtime based on per-byte estimates.
This moves the emulated simtime into emubd/kiwibd, and extends the idea
with both per-byte and per-op timing estimates for hopefully more
realistic results.
---
The problem is how NAND flash reads work.
Per-byte timing estimates are surprisingly accurate for NOR flash. There
is some overhead for sending the address, but it's mostly dominated by
bus cost (~20ns/B [1]).
NAND flash, on the otherhand, technically does support byte-level reads,
but first needs to read into 2KiB buffer. Surprisingly, these are pretty
close in cost (~19ns/B bus [2] vs ~12ns/B buffer [2]).
This close-ness makes modeling NAND flash difficult. If we set
read_size=1, we risk hiding the cost of small reads, which littlefs3 is
full of (rbyd lookups). If we set read_size=2048, we unfairly penalize
littlefs3 for the same reason.
---
The solution here is to expose both per-byte and per-op timing
estimates. This lets you model NAND reads using two data points:
^
| realtime --> ...............o
| : .....'''' :
| ...............:'''' ^ :
| :....''''' | :
| ..........:::::: simtime :
| .....:'''' :
|o....:::::.....: :
|: :
|: :
+:-----------------------------------------------------------:>
min read max read
Where:
bus_timing = 19ns
buffer_timing = 25us
buffer_size = 2KiB
erase_size = 128KiB
min_read = buffer_timing
max_read = (erase_size/buffer_size)*buffer_timing - buffer_timing
read_timing = min_read
readed_timing = ((max_read - min_read)/erase_size) + bus_timing
simtime = reads*read_timing + readed*readed_timing
(per-op) (per-byte)
This should correctly penalize small reads without complicating
emubd/kiwibd too much.
That's the idea anyways! It will take some use to understand if this is
a reasonable approach.
As a plus, this is a superset of the per-byte model, so both can be used
for realistic vs idealistic simulations (and to test the bus+buffer
model itself).
1: https://www.winbond.com/resource-files/W25Q256JV%20SPI%20RevQ%2002072025%20Plus.pdf
2: https://www.winbond.com/resource-files/W25N01GV%20Rev%20R%20070323.pdf
169 lines
5.1 KiB
C
169 lines
5.1 KiB
C
/*
|
|
* kiwibd - A lightweight variant of emubd, useful for emulating large
|
|
* disks backed by a file or in RAM.
|
|
*
|
|
* Unlike emubd, file-backed disks are _not_ mirrored in RAM. kiwibd has
|
|
* fewer features than emubd, prioritizing speed for benchmarking.
|
|
*
|
|
*
|
|
*/
|
|
#ifndef LFS3_KIWIBD_H
|
|
#define LFS3_KIWIBD_H
|
|
|
|
#include "lfs3.h"
|
|
#include "lfs3_util.h"
|
|
|
|
|
|
// Block device specific tracing
|
|
#ifndef LFS3_KIWIBD_TRACE
|
|
#ifdef LFS3_KIWIBD_YES_TRACE
|
|
#define LFS3_KIWIBD_TRACE(...) LFS3_TRACE(__VA_ARGS__)
|
|
#else
|
|
#define LFS3_KIWIBD_TRACE(...)
|
|
#endif
|
|
#endif
|
|
|
|
// Type for measuring read/program/erase operations
|
|
typedef uint64_t lfs3_kiwibd_io_t;
|
|
typedef int64_t lfs3_kiwibd_sio_t;
|
|
|
|
// Type for delays in nanoseconds
|
|
typedef uint64_t lfs3_kiwibd_ns_t;
|
|
typedef int64_t lfs3_kiwibd_sns_t;
|
|
|
|
// kiwibd config, this is required for testing
|
|
struct lfs3_kiwibd_cfg {
|
|
// Optional statically allocated buffer for the block device. Ignored
|
|
// if disk_path is provided.
|
|
void *buffer;
|
|
|
|
// 8-bit erase value to use for simulating erases. -1 simulates a noop
|
|
// erase, which is faster than simulating a fixed erase value. -2 emulates
|
|
// nor-masking, which is useful for testing other filesystems (littlefs
|
|
// does _not_ rely on this!).
|
|
int32_t erase_value;
|
|
|
|
// Simulated read transaction timing in nanoseconds, this is added
|
|
// to simtime each read call, ignoring the requested size
|
|
lfs3_kiwibd_ns_t reads_timing;
|
|
|
|
// Simulated prog transaction timing in nanoseconds, this is added
|
|
// to simtime each prog call, ignoring the requested size
|
|
lfs3_kiwibd_ns_t progs_timing;
|
|
|
|
// Simulated erase transaction timing in nanoseconds, this is added
|
|
// to simtime each erase call, ignoring the requested size
|
|
lfs3_kiwibd_ns_t erases_timing;
|
|
|
|
// Simulated read byte timing in nanoseconds, this is scaled by the
|
|
// requested size and added to simtime each read call.
|
|
lfs3_kiwibd_ns_t readed_timing;
|
|
|
|
// Simulated prog byte timing in nanoseconds, this is scaled by the
|
|
// requested size and added to simtime each prog call.
|
|
lfs3_kiwibd_ns_t progged_timing;
|
|
|
|
// Simulated erase byte timing in nanoseconds, this is scaled by the
|
|
// requested size and added to simtime each erase call.
|
|
lfs3_kiwibd_ns_t erased_timing;
|
|
|
|
// Artificial read transaction delay in nanoseconds, there is no
|
|
// purpose for this other than slowing down the simulation.
|
|
lfs3_kiwibd_ns_t read_sleep;
|
|
|
|
// Artificial prog transaction delay in nanoseconds, there is no
|
|
// purpose for this other than slowing down the simulation.
|
|
lfs3_kiwibd_ns_t prog_sleep;
|
|
|
|
// Artificial erase transaction delay in nanoseconds, there is no
|
|
// purpose for this other than slowing down the simulation.
|
|
lfs3_kiwibd_ns_t erase_sleep;
|
|
};
|
|
|
|
// kiwibd state
|
|
typedef struct lfs3_kiwibd {
|
|
// backing disk
|
|
int fd;
|
|
union {
|
|
uint8_t *scratch;
|
|
uint8_t *mem;
|
|
} u;
|
|
|
|
// amount read/progged/erased
|
|
lfs3_kiwibd_io_t reads;
|
|
lfs3_kiwibd_io_t progs;
|
|
lfs3_kiwibd_io_t erases;
|
|
lfs3_kiwibd_io_t readed;
|
|
lfs3_kiwibd_io_t progged;
|
|
lfs3_kiwibd_io_t erased;
|
|
|
|
const struct lfs3_kiwibd_cfg *cfg;
|
|
} lfs3_kiwibd_t;
|
|
|
|
|
|
/// Block device API ///
|
|
|
|
// Create a kiwibd using the geometry in lfs3_cfg
|
|
//
|
|
// If path is provided, kiwibd will use the file to back the block
|
|
// device, allowing emulation of block devices > available RAM.
|
|
//
|
|
int lfs3_kiwibd_create(const struct lfs3_cfg *cfg, const char *path);
|
|
int lfs3_kiwibd_createcfg(const struct lfs3_cfg *cfg, const char *path,
|
|
const struct lfs3_kiwibd_cfg *bdcfg);
|
|
|
|
// Clean up memory associated with block device
|
|
int lfs3_kiwibd_destroy(const struct lfs3_cfg *cfg);
|
|
|
|
// Read a block
|
|
int lfs3_kiwibd_read(const struct lfs3_cfg *cfg, lfs3_block_t block,
|
|
lfs3_off_t off, void *buffer, lfs3_size_t size);
|
|
|
|
// Program a block
|
|
//
|
|
// The block must have previously been erased.
|
|
int lfs3_kiwibd_prog(const struct lfs3_cfg *cfg, lfs3_block_t block,
|
|
lfs3_off_t off, const void *buffer, lfs3_size_t size);
|
|
|
|
// Erase a block
|
|
//
|
|
// A block must be erased before being programmed. The
|
|
// state of an erased block is undefined.
|
|
int lfs3_kiwibd_erase(const struct lfs3_cfg *cfg, lfs3_block_t block);
|
|
|
|
// Sync the block device
|
|
int lfs3_kiwibd_sync(const struct lfs3_cfg *cfg);
|
|
|
|
|
|
/// Additional kiwibd features ///
|
|
|
|
// Get total simulated runtime
|
|
lfs3_kiwibd_sns_t lfs3_kiwibd_simtime(const struct lfs3_cfg *cfg);
|
|
|
|
// Reset simulation counters
|
|
//
|
|
// You probably shouldn't call this, instead diff before/after simtimes
|
|
int lfs3_kiwibd_simreset(const struct lfs3_cfg *cfg);
|
|
|
|
// Get total number of read transactions
|
|
lfs3_kiwibd_sio_t lfs3_kiwibd_reads(const struct lfs3_cfg *cfg);
|
|
|
|
// Get total number of prog transactions
|
|
lfs3_kiwibd_sio_t lfs3_kiwibd_progs(const struct lfs3_cfg *cfg);
|
|
|
|
// Get total number of erase transactions
|
|
lfs3_kiwibd_sio_t lfs3_kiwibd_erases(const struct lfs3_cfg *cfg);
|
|
|
|
// Get total amount of bytes read
|
|
lfs3_kiwibd_sio_t lfs3_kiwibd_readed(const struct lfs3_cfg *cfg);
|
|
|
|
// Get total amount of bytes programmed
|
|
lfs3_kiwibd_sio_t lfs3_kiwibd_progged(const struct lfs3_cfg *cfg);
|
|
|
|
// Get total amount of bytes erased
|
|
lfs3_kiwibd_sio_t lfs3_kiwibd_erased(const struct lfs3_cfg *cfg);
|
|
|
|
|
|
#endif
|
|
|