[PATCH net-next 6/8] ice: reduce size of queue context fields
From: Jacob Keller
Date: Fri Oct 11 2024 - 14:53:22 EST
The ice_rlan_ctx and ice_tlan_ctx structures have some fields which are
intentionally sized larger necessary relative to the packed sizes the data
must fit into. This was done because the original ice_set_ctx() function
and its helpers did not correctly handle packing when the packed bits
straddled a byte. This is no longer the case with the use of the
<linux/packing.h> implementation.
Save some bytes in these structures by sizing the variables to the number
of bytes the actual bitpacked fields fit into.
There are a couple of gaps left in the structure, which is a result of the
fields being in the order they appear in the packed bit layout, but where
alignment forces some extra gaps. We could fix this, saving ~8 bytes from
each structure. However, these structures are not used heavily, and the
resulting savings is minimal:
$ bloat-o-meter ice-before-reorder.ko ice-after-reorder.ko
add/remove: 0/0 grow/shrink: 1/1 up/down: 26/-70 (-44)
Function old new delta
ice_vsi_cfg_txq 1873 1899 +26
ice_setup_rx_ctx.constprop 1529 1459 -70
Total: Before=1459555, After=1459511, chg -0.00%
Thus, the fields are left in the same order as the packed bit layout,
despite the gaps this causes.
Signed-off-by: Jacob Keller <jacob.e.keller@xxxxxxxxx>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@xxxxxxxxx>
---
drivers/net/ethernet/intel/ice/ice_lan_tx_rx.h | 32 ++++++++------------------
1 file changed, 10 insertions(+), 22 deletions(-)
diff --git a/drivers/net/ethernet/intel/ice/ice_lan_tx_rx.h b/drivers/net/ethernet/intel/ice/ice_lan_tx_rx.h
index 6c83e9d71c64..618cc39bd397 100644
--- a/drivers/net/ethernet/intel/ice/ice_lan_tx_rx.h
+++ b/drivers/net/ethernet/intel/ice/ice_lan_tx_rx.h
@@ -377,23 +377,17 @@ enum ice_rx_flex_desc_status_error_1_bits {
#define ICE_TX_DRBELL_Q_CTX_SIZE_DWORDS 5
#define GLTCLAN_CQ_CNTX(i, CQ) (GLTCLAN_CQ_CNTX0(CQ) + ((i) * 0x0800))
-/* RLAN Rx queue context data
- *
- * The sizes of the variables may be larger than needed due to crossing byte
- * boundaries. If we do not have the width of the variable set to the correct
- * size then we could end up shifting bits off the top of the variable when the
- * variable is at the top of a byte and crosses over into the next byte.
- */
+/* RLAN Rx queue context data */
struct ice_rlan_ctx {
u16 head;
- u16 cpuid; /* bigger than needed, see above for reason */
+ u8 cpuid;
#define ICE_RLAN_BASE_S 7
u64 base;
u16 qlen;
#define ICE_RLAN_CTX_DBUF_S 7
- u16 dbuf; /* bigger than needed, see above for reason */
+ u8 dbuf;
#define ICE_RLAN_CTX_HBUF_S 6
- u16 hbuf; /* bigger than needed, see above for reason */
+ u8 hbuf;
u8 dtype;
u8 dsize;
u8 crcstrip;
@@ -401,12 +395,12 @@ struct ice_rlan_ctx {
u8 hsplit_0;
u8 hsplit_1;
u8 showiv;
- u32 rxmax; /* bigger than needed, see above for reason */
+ u16 rxmax;
u8 tphrdesc_ena;
u8 tphwdesc_ena;
u8 tphdata_ena;
u8 tphhead_ena;
- u16 lrxqthresh; /* bigger than needed, see above for reason */
+ u8 lrxqthresh;
u8 prefena; /* NOTE: normally must be set to 1 at init */
};
@@ -539,18 +533,12 @@ enum ice_tx_ctx_desc_eipt_offload {
#define ICE_TXQ_CTX_SZ 22
-/* Tx queue context data
- *
- * The sizes of the variables may be larger than needed due to crossing byte
- * boundaries. If we do not have the width of the variable set to the correct
- * size then we could end up shifting bits off the top of the variable when the
- * variable is at the top of a byte and crosses over into the next byte.
- */
+/* Tx queue context data */
struct ice_tlan_ctx {
#define ICE_TLAN_CTX_BASE_S 7
u64 base; /* base is defined in 128-byte units */
u8 port_num;
- u16 cgd_num; /* bigger than needed, see above for reason */
+ u8 cgd_num;
u8 pf_num;
u16 vmvf_num;
u8 vmvf_type;
@@ -561,7 +549,7 @@ struct ice_tlan_ctx {
u8 tsyn_ena;
u8 internal_usage_flag;
u8 alt_vlan;
- u16 cpuid; /* bigger than needed, see above for reason */
+ u8 cpuid;
u8 wb_mode;
u8 tphrd_desc;
u8 tphrd;
@@ -570,7 +558,7 @@ struct ice_tlan_ctx {
u16 qnum_in_func;
u8 itr_notification_mode;
u8 adjust_prof_id;
- u32 qlen; /* bigger than needed, see above for reason */
+ u16 qlen;
u8 quanta_prof_idx;
u8 tso_ena;
u16 tso_qnum;
--
2.47.0.265.g4ca455297942