xen/arm: Clean-up & fixes in boot/mm code
v2 cover xen/arm: Clean-up & fixes in boot/mm code
Hi all,
This is the second part of the boot/memory rework for Xen on Arm. This part contains mostly clean-up & fixes found during the rework.
The first part of the rework is "xen/arm: TLB flush helpers rework" [1].
For convenience, I provided a branch with all the patches applied based on staging:
git://xenbits.xen.org/people/julieng/xen-unstable.git branch mm/part2/v2
Cheers,
[1] https://lists.xenproject.org/archives/html/xen-devel/2019-05/msg01109.html
Julien Grall (19):
xen/const: Extend the existing macro BIT to take a suffix in parameter xen/arm: Rename SCTLR_* defines and remove unused one xen/arm: processor: Use BIT(.., UL) instead of _AC(1, U) in SCTLR_
defines
xen/arm: Rework HSCTLR_BASE
xen/arm: Remove parameter cpuid from start_xen
xen/arm: Rework secondary_start prototype
xen/arm64: head: Remove unnecessary comment
xen/arm64: head: Move earlyprintk messages in .rodata.str xen/arm64: head: Correctly report the HW CPU ID
xen/arm32: head: Correctly report the HW CPU ID
xen/arm32: head: Don't set MAIR0 and MAIR1
xen/arm32: head: Always zero r3 before update a page-table entry xen/arm32: mm: Avoid to zero and clean cache for CPU0 domheap xen/arm32: mm: Avoid cleaning the cache for secondary CPUs page-tables xen/arm: mm: Introduce DEFINE_PAGE_TABLE{,S} and use it xen/arm: mm: Protect Xen page-table update with a spinlock xen/arm: mm: Initialize page-tables earlier
xen/arm: mm: Check start is always before end in {destroy,
modify}_xen_mappings
xen/arm: Pair call to set_fixmap with call to clear_fixmap in
copy_from_paddr
xen/arch/arm/arm32/head.S | 34 ++++---------- xen/arch/arm/arm32/insn.c | 2 +- xen/arch/arm/arm64/head.S | 40 +++++------------ xen/arch/arm/arm64/insn.c | 18 ++++---- xen/arch/arm/gic-v3-its.c | 13 +++--- xen/arch/arm/gic-v3-lpi.c | 4 +- xen/arch/arm/guest_walk.c | 10 ++--- xen/arch/arm/kernel.c | 3 +- xen/arch/arm/mm.c | 62 +++++++++++++------------- xen/arch/arm/setup.c | 7 ++- xen/arch/arm/smpboot.c | 4 +- xen/arch/arm/traps.c | 6 +-- xen/arch/arm/vgic-v3-its.c | 12 ++--- xen/arch/arm/vgic-v3.c | 2 +- xen/arch/arm/vgic.c | 2 +- xen/drivers/char/meson-uart.c | 16 +++---- xen/drivers/char/mvebu-uart.c | 34 +++++++------- xen/include/asm-arm/asm_defns.h | 5 +++ xen/include/asm-arm/bitops.h | 2 - xen/include/asm-arm/gic_v3_defs.h | 4 +- xen/include/asm-arm/gic_v3_its.h | 10 ++--- xen/include/asm-arm/p2m.h | 4 +- xen/include/asm-arm/processor.h | 93 ++++++++++++++++++++++++++++++--------- xen/include/xen/const.h | 2 + 24 files changed, 201 insertions(+), 188 deletions(-)
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 1/19 xen/const: Extend the existing macro BIT to take a suffix in parameter
Arm currently provides two macro BIT and BIT_ULL that are only usable in C and return respectively unsigned long and unsigned long long.
Extending the macros to deal with assembly would be a nice benefits as it could replace the common pattern to define fields (AC(1, sfx) << X) easier to read.
Rather than extending the two macros, it was decided to drop BIT_ULL() and extend the macro BIT() to take a suffix (e.g U, UL, ULL) in parameter. This would allow to use different suffix without having to define new macros.
The new extend macro is now moved in include/xen/const.h so it can be used by anyone in Xen and also avoid to include bitops.h in assembly code.
---
Changes in v2:
- Replace "xen/const: Introduce _BITUL and _BITULL"
--- xen/arch/arm/arm32/insn.c | 2 +- xen/arch/arm/arm64/insn.c | 18 +++++++++--------- xen/arch/arm/gic-v3-its.c | 13 +++++++------ xen/arch/arm/gic-v3-lpi.c | 4 ++-- xen/arch/arm/guest_walk.c | 8 ++++---- xen/arch/arm/vgic-v3-its.c | 12 ++++++------ xen/arch/arm/vgic-v3.c | 2 +- xen/arch/arm/vgic.c | 2 +- xen/drivers/char/meson-uart.c | 16 ++++++++-------- xen/drivers/char/mvebu-uart.c | 34 +++++++++++++++++----------------- xen/include/asm-arm/bitops.h | 2 -- xen/include/asm-arm/gic_v3_defs.h | 4 ++-- xen/include/asm-arm/gic_v3_its.h | 10 +++++----- xen/include/xen/const.h | 2 ++ 14 files changed, 65 insertions(+), 64 deletions(-)
xen/arch/arm/arm32/insn.c +1 -1
@@ -58,7 +58,7 @@ int32_t aarch32_get_branch_offset(uint32_t insn)* Check the imm signed bit. If the imm is a negative value, we * have to extend the imm to a full 32 bit negative value. */ -if ( imm &BIT (23 ) ) +if ( imm &BIT (23 ,UL ) ) imm |=GENMASK (31 ,24 );return (int32_t )(imm <<2 );
xen/arch/arm/arm64/insn.c +9 -9
@@ -45,40 +45,40 @@ static int __kprobes aarch64_get_imm_shift_mask(enum aarch64_insn_imm_type type,switch (type) {case AARCH64_INSN_IMM_26 : - mask =BIT (26 ) -1 ; + mask =BIT (26 ,UL ) -1 ; shift =0 ;break ;case AARCH64_INSN_IMM_19 : - mask =BIT (19 ) -1 ; + mask =BIT (19 ,UL ) -1 ; shift =5 ;break ;case AARCH64_INSN_IMM_16 : - mask =BIT (16 ) -1 ; + mask =BIT (16 ,UL ) -1 ; shift =5 ;break ;case AARCH64_INSN_IMM_14 : - mask =BIT (14 ) -1 ; + mask =BIT (14 ,UL ) -1 ; shift =5 ;break ;case AARCH64_INSN_IMM_12 : - mask =BIT (12 ) -1 ; + mask =BIT (12 ,UL ) -1 ; shift =10 ;break ;case AARCH64_INSN_IMM_9 : - mask =BIT (9 ) -1 ; + mask =BIT (9 ,UL ) -1 ; shift =12 ;break ;case AARCH64_INSN_IMM_7 : - mask =BIT (7 ) -1 ; + mask =BIT (7 ,UL ) -1 ; shift =15 ;break ;case AARCH64_INSN_IMM_6 :case AARCH64_INSN_IMM_S : - mask =BIT (6 ) -1 ; + mask =BIT (6 ,UL ) -1 ; shift =10 ;break ;case AARCH64_INSN_IMM_R : - mask =BIT (6 ) -1 ; + mask =BIT (6 ,UL ) -1 ; shift =16 ;break ;default :
xen/arch/arm/gic-v3-its.c +7 -6
@@ -363,11 +363,12 @@ static int its_map_baser(void __iomem *basereg, uint64_t regc,* attributes), retrying if necessary. */ retry: - table_size =ROUNDUP (nr_items * entry_size,BIT (BASER_PAGE_BITS (pagesz))); + table_size =ROUNDUP (nr_items * entry_size, +BIT (BASER_PAGE_BITS (pagesz),UL ));/* The BASE registers support at most 256 pages. */ table_size =min (table_size,256U <<BASER_PAGE_BITS (pagesz)); - buffer =_xzalloc (table_size,BIT (BASER_PAGE_BITS (pagesz))); + buffer =_xzalloc (table_size,BIT (BASER_PAGE_BITS (pagesz),UL ));if ( !buffer )return -ENOMEM ;
@@ -483,7 +484,7 @@ static int gicv3_its_init_single_its(struct host_its *hw_its)case GITS_BASER_TYPE_NONE :continue ;case GITS_BASER_TYPE_DEVICE : - ret =its_map_baser (basereg, reg,BIT (hw_its->devid_bits)); + ret =its_map_baser (basereg, reg,BIT (hw_its->devid_bits,UL ));if ( ret )return ret;break ;
@@ -635,7 +636,7 @@ int gicv3_its_map_guest_device(struct domain *d,return ret;/* Sanitise the provided hardware values against the host ITS. */ -if ( host_devid >=BIT (hw_its->devid_bits) ) +if ( host_devid >=BIT (hw_its->devid_bits,UL ) )return -EINVAL ;/*
@@ -645,10 +646,10 @@ int gicv3_its_map_guest_device(struct domain *d,* TODO: Investigate if the number of events can be limited to smaller * values if the guest does not require that many. */ - nr_events =BIT (fls (nr_events -1 )); + nr_events =BIT (fls (nr_events -1 ),UL );if ( nr_events <LPI_BLOCK ) nr_events =LPI_BLOCK ; -if ( nr_events >=BIT (hw_its->evid_bits) ) +if ( nr_events >=BIT (hw_its->evid_bits,UL ) )return -EINVAL ;/* check for already existing mappings */
xen/arch/arm/gic-v3-lpi.c +2 -2
@@ -392,14 +392,14 @@ int gicv3_lpi_init_host_lpis(unsigned int host_lpi_bits)printk (XENLOG_WARNING "WARNING: max_lpi_bits must be between 14 and 32, adjusting.\n" ); max_lpi_bits =max (max_lpi_bits,14U ); - lpi_data.max_host_lpi_ids =BIT (min (host_lpi_bits, max_lpi_bits)); + lpi_data.max_host_lpi_ids =BIT (min (host_lpi_bits, max_lpi_bits),UL );/* * Warn if the number of LPIs are quite high, as the user might not want * to waste megabytes of memory for a mostly empty table. * It's very unlikely that we need more than 24 bits worth of LPIs. */ -if ( lpi_data.max_host_lpi_ids >BIT (24 ) ) +if ( lpi_data.max_host_lpi_ids >BIT (24 ,UL ) )warning_add ("Using high number of LPIs, limit memory usage with max_lpi_bits\n" );spin_lock_init (&lpi_data.host_lpis_lock);
xen/arch/arm/guest_walk.c +4 -4
@@ -327,8 +327,8 @@ static unsigned int get_top_bit(struct domain *d, vaddr_t gva, register_t tcr) topbit =31 ; else { -if ( ((gva &BIT_ULL (55 )) && (tcr &TCR_EL1_TBI1 )) || - (!(gva &BIT_ULL (55 )) && (tcr &TCR_EL1_TBI0 )) ) +if ( ((gva &BIT (55 ,ULL )) && (tcr &TCR_EL1_TBI1 )) || + (!(gva &BIT (55 ,ULL )) && (tcr &TCR_EL1_TBI0 )) ) topbit =55 ;else topbit =63 ;
@@ -419,7 +419,7 @@ static bool guest_walk_ld(const struct vcpu *v, {/* Select the TTBR(0|1)_EL1 that will be used for address translation. */ -if ( (gva &BIT_ULL (topbit)) ==0 ) +if ( (gva &BIT (topbit,ULL )) ==0 ) { input_size =64 - t0_sz;
@@ -554,7 +554,7 @@ static bool guest_walk_ld(const struct vcpu *v,* inherited by page table attributes (ARM DDI 0487B.a J1-5928). */ xn_table |= pte.pt.xnt;/* Execute-Never */ - ro_table |= pte.pt.apt &BIT (1 );/* Read-Only */ + ro_table |= pte.pt.apt &BIT (1 ,UL );/* Read-Only */ /* Compute the base address of the next level translation table. */ mask =GENMASK_ULL (47 , grainsizes[gran]);
xen/arch/arm/vgic-v3-its.c +6 -6
@@ -97,7 +97,7 @@ typedef uint16_t coll_table_entry_t;*/ typedef uint64_t dev_table_entry_t ;#define DEV_TABLE_ITT_ADDR (x) ((x) & GENMASK(51, 8)) -#define DEV_TABLE_ITT_SIZE (x) (BIT(((x) & GENMASK(4, 0)) + 1)) +#define DEV_TABLE_ITT_SIZE (x) (BIT(((x) & GENMASK(4, 0)) + 1, UL))#define DEV_TABLE_ENTRY (addr, bits) \ (((addr) & GENMASK(51, 8)) | (((bits) - 1) & GENMASK(4, 0)))
@@ -111,7 +111,7 @@ typedef uint64_t dev_table_entry_t;*/ static paddr_t get_baser_phys_addr (uint64_t reg) { -if ( reg &BIT (9 ) ) +if ( reg &BIT (9 ,UL ) )return (reg &GENMASK (47 ,16 )) | ((reg &GENMASK (15 ,12 )) <<36 );else
@@ -125,7 +125,7 @@ static int its_set_collection(struct virt_its *its, uint16_t collid,paddr_t addr =get_baser_phys_addr (its->baser_coll);/* The collection table entry must be able to store a VCPU ID. */ -BUILD_BUG_ON (BIT (sizeof (coll_table_entry_t) *8 ) <MAX_VIRT_CPUS ); +BUILD_BUG_ON (BIT (sizeof (coll_table_entry_t) *8 ,UL ) <MAX_VIRT_CPUS );ASSERT (spin_is_locked (&its->its_lock));
@@ -690,7 +690,7 @@ static int its_handle_mapd(struct virt_its *its, uint64_t *cmdptr)*/ ret =gicv3_its_map_guest_device (its->d, its->doorbell_address, devid, its->doorbell_address, devid, -BIT (size), valid); +BIT (size,UL ), valid);if ( ret && valid )return ret; }
@@ -1356,8 +1356,8 @@ static int vgic_v3_its_mmio_write(struct vcpu *v, mmio_info_t *info,if ( reg &GITS_VALID_BIT ) { its->max_devices =its_baser_nr_entries (reg); -if ( its->max_devices >BIT (its->devid_bits) ) - its->max_devices =BIT (its->devid_bits); +if ( its->max_devices >BIT (its->devid_bits,UL ) ) + its->max_devices =BIT (its->devid_bits,UL ); }else its->max_devices =0 ;
xen/arch/arm/vgic-v3.c +1 -1
@@ -442,7 +442,7 @@ static uint64_t sanitize_pendbaser(uint64_t reg)static void vgic_vcpu_enable_lpis (struct vcpu *v) {uint64_t reg = v->domain->arch.vgic.rdist_propbase; -unsigned int nr_lpis =BIT ((reg &0x1f ) +1 ); +unsigned int nr_lpis =BIT ((reg &0x1f ) +1 ,UL );/* rdists_enabled is protected by the domain lock. */ ASSERT (spin_is_locked (&v->domain->arch.vgic.lock));
xen/arch/arm/vgic.c +1 -1
@@ -65,7 +65,7 @@ struct vgic_irq_rank *vgic_rank_irq(struct vcpu *v, unsigned int irq)void vgic_init_pending_irq (struct pending_irq *p,unsigned int virq) {/* The lpi_vcpu_id field must be big enough to hold a VCPU ID. */ -BUILD_BUG_ON (BIT (sizeof (p->lpi_vcpu_id) *8 ) <MAX_VIRT_CPUS ); +BUILD_BUG_ON (BIT (sizeof (p->lpi_vcpu_id) *8 ,UL ) <MAX_VIRT_CPUS );memset (p,0 ,sizeof (*p));INIT_LIST_HEAD (&p->inflight);
xen/drivers/char/meson-uart.c +8 -8
@@ -31,16 +31,16 @@#define AML_UART_MISC_REG 0x10/* UART_CONTROL bits */ -#define AML_UART_TX_RST BIT(22) -#define AML_UART_RX_RST BIT(23) -#define AML_UART_CLEAR_ERR BIT(24) -#define AML_UART_RX_INT_EN BIT(27) -#define AML_UART_TX_INT_EN BIT(28) +#define AML_UART_TX_RST BIT(22, UL) +#define AML_UART_RX_RST BIT(23, UL) +#define AML_UART_CLEAR_ERR BIT(24, UL) +#define AML_UART_RX_INT_EN BIT(27, UL) +#define AML_UART_TX_INT_EN BIT(28, UL)/* UART_STATUS bits */ -#define AML_UART_RX_FIFO_EMPTY BIT(20) -#define AML_UART_TX_FIFO_FULL BIT(21) -#define AML_UART_TX_FIFO_EMPTY BIT(22) +#define AML_UART_RX_FIFO_EMPTY BIT(20, UL) +#define AML_UART_TX_FIFO_FULL BIT(21, UL) +#define AML_UART_TX_FIFO_EMPTY BIT(22, UL)#define AML_UART_TX_CNT_MASK GENMASK(14, 8)/* AML_UART_MISC bits */
xen/drivers/char/mvebu-uart.c +17 -17
@@ -29,27 +29,27 @@#define UART_TX_REG 0x04#define UART_CTRL_REG 0x08 -#define CTRL_TXFIFO_RST BIT(15) -#define CTRL_RXFIFO_RST BIT(14) -#define CTRL_TX_RDY_INT BIT(5) -#define CTRL_RX_RDY_INT BIT(4) -#define CTRL_BRK_DET_INT BIT(3) -#define CTRL_FRM_ERR_INT BIT(2) -#define CTRL_PAR_ERR_INT BIT(1) -#define CTRL_OVR_ERR_INT BIT(0) +#define CTRL_TXFIFO_RST BIT(15, UL) +#define CTRL_RXFIFO_RST BIT(14, UL) +#define CTRL_TX_RDY_INT BIT(5, UL) +#define CTRL_RX_RDY_INT BIT(4, UL) +#define CTRL_BRK_DET_INT BIT(3, UL) +#define CTRL_FRM_ERR_INT BIT(2, UL) +#define CTRL_PAR_ERR_INT BIT(1, UL) +#define CTRL_OVR_ERR_INT BIT(0, UL)#define CTRL_ERR_INT (CTRL_BRK_DET_INT | CTRL_FRM_ERR_INT | \ CTRL_PAR_ERR_INT | CTRL_OVR_ERR_INT)#define UART_STATUS_REG 0x0c -#define STATUS_TXFIFO_EMP BIT(13) -#define STATUS_TXFIFO_FUL BIT(11) -#define STATUS_TXFIFO_HFL BIT(10) -#define STATUS_TX_RDY BIT(5) -#define STATUS_RX_RDY BIT(4) -#define STATUS_BRK_DET BIT(3) -#define STATUS_FRM_ERR BIT(2) -#define STATUS_PAR_ERR BIT(1) -#define STATUS_OVR_ERR BIT(0) +#define STATUS_TXFIFO_EMP BIT(13, UL) +#define STATUS_TXFIFO_FUL BIT(11, UL) +#define STATUS_TXFIFO_HFL BIT(10, UL) +#define STATUS_TX_RDY BIT(5, UL) +#define STATUS_RX_RDY BIT(4, UL) +#define STATUS_BRK_DET BIT(3, UL) +#define STATUS_FRM_ERR BIT(2, UL) +#define STATUS_PAR_ERR BIT(1, UL) +#define STATUS_OVR_ERR BIT(0, UL)#define STATUS_BRK_ERR (STATUS_BRK_DET | STATUS_FRM_ERR | \ STATUS_PAR_ERR | STATUS_OVR_ERR)
xen/include/asm-arm/bitops.h +0 -2
@@ -21,10 +21,8 @@#define __clear_bit (n,p) clear_bit(n,p)#define BITS_PER_WORD 32 -#define BIT (nr) (1UL << (nr))#define BIT_MASK (nr) (1UL << ((nr) % BITS_PER_WORD))#define BIT_WORD (nr) ((nr) / BITS_PER_WORD) -#define BIT_ULL (nr) (1ULL << (nr))#define BITS_PER_BYTE 8#define ADDR (*(volatile int *) addr)
xen/include/asm-arm/gic_v3_defs.h +2 -2
@@ -149,9 +149,9 @@ (7UL << GICR_PENDBASER_INNER_CACHEABILITY_SHIFT)#define GICR_PENDBASER_OUTER_CACHEABILITY_MASK \ (7UL << GICR_PENDBASER_OUTER_CACHEABILITY_SHIFT) -#define GICR_PENDBASER_PTZ BIT(62) +#define GICR_PENDBASER_PTZ BIT(62, UL)#define GICR_PENDBASER_RES0_MASK \ - (BIT(63) | GENMASK(61, 59) | GENMASK(55, 52) | \ + (BIT(63, UL) | GENMASK(61, 59) | GENMASK(55, 52) | \ GENMASK(15, 12) | GENMASK(6, 0))#define DEFAULT_PMR_VALUE 0xff
xen/include/asm-arm/gic_v3_its.h +5 -5
@@ -38,12 +38,12 @@#define GITS_PIDR2 GICR_PIDR2/* Register bits */ -#define GITS_VALID_BIT BIT(63) +#define GITS_VALID_BIT BIT(63, UL) -#define GITS_CTLR_QUIESCENT BIT(31) -#define GITS_CTLR_ENABLE BIT(0) +#define GITS_CTLR_QUIESCENT BIT(31, UL) +#define GITS_CTLR_ENABLE BIT(0, UL) -#define GITS_TYPER_PTA BIT(19) +#define GITS_TYPER_PTA BIT(19, UL)#define GITS_TYPER_DEVIDS_SHIFT 13#define GITS_TYPER_DEVIDS_MASK (0x1fUL << GITS_TYPER_DEVIDS_SHIFT)#define GITS_TYPER_DEVICE_ID_BITS (r) (((r & GITS_TYPER_DEVIDS_MASK) >> \
@@ -60,7 +60,7 @@ GITS_TYPER_ITT_SIZE_SHIFT) + 1)#define GITS_TYPER_PHYSICAL (1U << 0) -#define GITS_BASER_INDIRECT BIT(62) +#define GITS_BASER_INDIRECT BIT(62, UL)#define GITS_BASER_INNER_CACHEABILITY_SHIFT 59#define GITS_BASER_TYPE_SHIFT 56#define GITS_BASER_TYPE_MASK (7ULL << GITS_BASER_TYPE_SHIFT)
xen/include/xen/const.h +2 -0
@@ -21,4 +21,6 @@#define _AT (T ,X ) ((T)(X))#endif +#define BIT (pos, sfx) (_AC(1, sfx) << (pos)) +#endif /* __XEN_CONST_H__ */
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 01/19] xen/const: Extend the existing macro BIT to take a suffix in parameter
Quoted 13 lines
Quoted 1 line
Quoted 1 line · On 14.05.19 at 14:24, <[email protected]> wrote:
Arm currently provides two macro BIT and
BIT_ULLthat are only usable in C and return respectively unsigned long and unsigned long long.Extending the macros to deal with assembly would be a nice benefits as it could replace the common pattern to define fields (
AC(1, sfx)<< X) easier to read.Rather than extending the two macros, it was decided to drop
BIT_ULL()and extend the macroBIT()to take a suffix (e.g U, UL, ULL) in parameter. This would allow to use different suffix without having to define new macros.
Quoted 5 lines
The new extend macro is now moved in
include/xen/const.hso it can be used by anyone in Xen and also avoid to includebitops.hin assembly code.Signed-off-by Julien Grall <[email protected]>
Thanks much for going this route! For the single line addition to common code
Jan
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 01/19] xen/const: Extend the existing macro BIT to take a suffix in parameter
In reply to Julien GrallQuoted 12 lines · Julien Grall
Arm currently provides two macro BIT and
BIT_ULLthat are only usable in C and return respectively unsigned long and unsigned long long.Extending the macros to deal with assembly would be a nice benefits as it could replace the common pattern to define fields (
AC(1, sfx)<< X) easier to read.Rather than extending the two macros, it was decided to drop
BIT_ULL()and extend the macroBIT()to take a suffix (e.g U, UL, ULL) in parameter. This would allow to use different suffix without having to define new macros.
Quoted 5 lines
The new extend macro is now moved in
include/xen/const.hso it can be used by anyone in Xen and also avoid to includebitops.hin assembly code.Signed-off-by Julien Grall <[email protected]>
Quoted 421 lines
---
Changes in v2:
- Replace "xen/const: Introduce_BITULand_BITULL"--- xen/arch/arm/arm32/insn.c | 2 +- xen/arch/arm/arm64/insn.c | 18 +++++++++--------- xen/arch/arm/gic-v3-its.c | 13 +++++++------ xen/arch/arm/gic-v3-lpi.c | 4 ++-- xen/arch/arm/guest_walk.c | 8 ++++---- xen/arch/arm/vgic-v3-its.c | 12 ++++++------ xen/arch/arm/vgic-v3.c | 2 +- xen/arch/arm/vgic.c | 2 +- xen/drivers/char/meson-uart.c | 16 ++++++++-------- xen/drivers/char/mvebu-uart.c | 34 +++++++++++++++++----------------- xen/include/asm-arm/bitops.h | 2 -- xen/include/asm-arm/gic_v3_defs.h | 4 ++-- xen/include/asm-arm/gic_v3_its.h | 10 +++++----- xen/include/xen/const.h | 2 ++ 14 files changed, 65 insertions(+), 64 deletions(-)xen/arch/arm/arm32/insn.c +1
-1@@ -58,7 +58,7 @@ int32_t aarch32_get_branch_offset(uint32_t insn)* Check the imm signed bit. If the imm is a negative value, we * have to extend the imm to a full 32 bit negative value. */ -if ( imm &BIT (23 ) ) +if ( imm &BIT (23 ,UL ) ) imm |=GENMASK (31 ,24 );return (int32_t )(imm <<2 );xen/arch/arm/arm64/insn.c +9
-9@@ -45,40 +45,40 @@ static int __kprobes aarch64_get_imm_shift_mask(enum aarch64_insn_imm_type type,switch (type) {case AARCH64_INSN_IMM_26 : - mask =BIT (26 ) -1 ; + mask =BIT (26 ,UL ) -1 ; shift =0 ;break ;case AARCH64_INSN_IMM_19 : - mask =BIT (19 ) -1 ; + mask =BIT (19 ,UL ) -1 ; shift =5 ;break ;case AARCH64_INSN_IMM_16 : - mask =BIT (16 ) -1 ; + mask =BIT (16 ,UL ) -1 ; shift =5 ;break ;case AARCH64_INSN_IMM_14 : - mask =BIT (14 ) -1 ; + mask =BIT (14 ,UL ) -1 ; shift =5 ;break ;case AARCH64_INSN_IMM_12 : - mask =BIT (12 ) -1 ; + mask =BIT (12 ,UL ) -1 ; shift =10 ;break ;case AARCH64_INSN_IMM_9 : - mask =BIT (9 ) -1 ; + mask =BIT (9 ,UL ) -1 ; shift =12 ;break ;case AARCH64_INSN_IMM_7 : - mask =BIT (7 ) -1 ; + mask =BIT (7 ,UL ) -1 ; shift =15 ;break ;case AARCH64_INSN_IMM_6 :case AARCH64_INSN_IMM_S : - mask =BIT (6 ) -1 ; + mask =BIT (6 ,UL ) -1 ; shift =10 ;break ;case AARCH64_INSN_IMM_R : - mask =BIT (6 ) -1 ; + mask =BIT (6 ,UL ) -1 ; shift =16 ;break ;default :xen/arch/arm/gic-v3-its.c +7
-6@@ -363,11 +363,12 @@ static int its_map_baser(void __iomem *basereg, uint64_t regc,* attributes), retrying if necessary. */ retry: - table_size =ROUNDUP (nr_items * entry_size,BIT (BASER_PAGE_BITS (pagesz))); + table_size =ROUNDUP (nr_items * entry_size, +BIT (BASER_PAGE_BITS (pagesz),UL ));/* The BASE registers support at most 256 pages. */ table_size =min (table_size,256U <<BASER_PAGE_BITS (pagesz)); - buffer =_xzalloc (table_size,BIT (BASER_PAGE_BITS (pagesz))); + buffer =_xzalloc (table_size,BIT (BASER_PAGE_BITS (pagesz),UL ));if ( !buffer )return -ENOMEM ;@@ -483,7 +484,7 @@ static int gicv3_its_init_single_its(struct host_its *hw_its)case GITS_BASER_TYPE_NONE :continue ;case GITS_BASER_TYPE_DEVICE : - ret =its_map_baser (basereg, reg,BIT (hw_its->devid_bits)); + ret =its_map_baser (basereg, reg,BIT (hw_its->devid_bits,UL ));if ( ret )return ret;break ;@@ -635,7 +636,7 @@ int gicv3_its_map_guest_device(struct domain *d,return ret;/* Sanitise the provided hardware values against the host ITS. */ -if ( host_devid >=BIT (hw_its->devid_bits) ) +if ( host_devid >=BIT (hw_its->devid_bits,UL ) )return -EINVAL ;/* @@ -645,10 +646,10 @@ int gicv3_its_map_guest_device(struct domain *d,* TODO: Investigate if the number of events can be limited to smaller * values if the guest does not require that many. */ - nr_events =BIT (fls (nr_events -1 )); + nr_events =BIT (fls (nr_events -1 ),UL );if ( nr_events <LPI_BLOCK ) nr_events =LPI_BLOCK ; -if ( nr_events >=BIT (hw_its->evid_bits) ) +if ( nr_events >=BIT (hw_its->evid_bits,UL ) )return -EINVAL ;/* check for already existing mappings */ xen/arch/arm/gic-v3-lpi.c +2
-2@@ -392,14 +392,14 @@ int gicv3_lpi_init_host_lpis(unsigned int host_lpi_bits)printk (XENLOG_WARNING "WARNING: max_lpi_bits must be between 14 and 32, adjusting.\n" ); max_lpi_bits =max (max_lpi_bits,14U ); - lpi_data.max_host_lpi_ids =BIT (min (host_lpi_bits, max_lpi_bits)); + lpi_data.max_host_lpi_ids =BIT (min (host_lpi_bits, max_lpi_bits),UL );/* * Warn if the number of LPIs are quite high, as the user might not want * to waste megabytes of memory for a mostly empty table. * It's very unlikely that we need more than 24 bits worth of LPIs. */ -if ( lpi_data.max_host_lpi_ids >BIT (24 ) ) +if ( lpi_data.max_host_lpi_ids >BIT (24 ,UL ) )warning_add ("Using high number of LPIs, limit memory usage with max_lpi_bits\n" );spin_lock_init (&lpi_data.host_lpis_lock);xen/arch/arm/guest_walk.c +4
-4@@ -327,8 +327,8 @@ static unsigned int get_top_bit(struct domain *d, vaddr_t gva, register_t tcr) topbit =31 ; else { -if ( ((gva &BIT_ULL (55 )) && (tcr &TCR_EL1_TBI1 )) || - (!(gva &BIT_ULL (55 )) && (tcr &TCR_EL1_TBI0 )) ) +if ( ((gva &BIT (55 ,ULL )) && (tcr &TCR_EL1_TBI1 )) || + (!(gva &BIT (55 ,ULL )) && (tcr &TCR_EL1_TBI0 )) ) topbit =55 ;else topbit =63 ;@@ -419,7 +419,7 @@ static bool guest_walk_ld(const struct vcpu *v, {/* Select the TTBR(0|1)_EL1 that will be used for address translation. */ -if ( (gva &BIT_ULL (topbit)) ==0 ) +if ( (gva &BIT (topbit,ULL )) ==0 ) { input_size =64 - t0_sz;@@ -554,7 +554,7 @@ static bool guest_walk_ld(const struct vcpu *v,* inherited by page table attributes (ARM DDI 0487B.a J1-5928). */ xn_table |= pte.pt.xnt;/* Execute-Never */ - ro_table |= pte.pt.apt &BIT (1 );/* Read-Only */ + ro_table |= pte.pt.apt &BIT (1 ,UL );/* Read-Only */ /* Compute the base address of the next level translation table. */ mask =GENMASK_ULL (47 , grainsizes[gran]);xen/arch/arm/vgic-v3-its.c +6
-6@@ -97,7 +97,7 @@ typedef uint16_t coll_table_entry_t;*/ typedef uint64_t dev_table_entry_t ;#define DEV_TABLE_ITT_ADDR (x) ((x) & GENMASK(51, 8)) -#define DEV_TABLE_ITT_SIZE (x) (BIT(((x) & GENMASK(4, 0)) + 1)) +#define DEV_TABLE_ITT_SIZE (x) (BIT(((x) & GENMASK(4, 0)) + 1, UL))#define DEV_TABLE_ENTRY (addr, bits) \ (((addr) & GENMASK(51, 8)) | (((bits) - 1) & GENMASK(4, 0)))@@ -111,7 +111,7 @@ typedef uint64_t dev_table_entry_t;*/ static paddr_t get_baser_phys_addr (uint64_t reg) { -if ( reg &BIT (9 ) ) +if ( reg &BIT (9 ,UL ) )return (reg &GENMASK (47 ,16 )) | ((reg &GENMASK (15 ,12 )) <<36 );else @@ -125,7 +125,7 @@ static int its_set_collection(struct virt_its *its, uint16_t collid,paddr_t addr =get_baser_phys_addr (its->baser_coll);/* The collection table entry must be able to store a VCPU ID. */ -BUILD_BUG_ON (BIT (sizeof (coll_table_entry_t) *8 ) <MAX_VIRT_CPUS ); +BUILD_BUG_ON (BIT (sizeof (coll_table_entry_t) *8 ,UL ) <MAX_VIRT_CPUS );ASSERT (spin_is_locked (&its->its_lock));@@ -690,7 +690,7 @@ static int its_handle_mapd(struct virt_its *its, uint64_t *cmdptr)*/ ret =gicv3_its_map_guest_device (its->d, its->doorbell_address, devid, its->doorbell_address, devid, -BIT (size), valid); +BIT (size,UL ), valid);if ( ret && valid )return ret; }@@ -1356,8 +1356,8 @@ static int vgic_v3_its_mmio_write(struct vcpu *v, mmio_info_t *info,if ( reg &GITS_VALID_BIT ) { its->max_devices =its_baser_nr_entries (reg); -if ( its->max_devices >BIT (its->devid_bits) ) - its->max_devices =BIT (its->devid_bits); +if ( its->max_devices >BIT (its->devid_bits,UL ) ) + its->max_devices =BIT (its->devid_bits,UL ); }else its->max_devices =0 ;xen/arch/arm/vgic-v3.c +1
-1@@ -442,7 +442,7 @@ static uint64_t sanitize_pendbaser(uint64_t reg)static void vgic_vcpu_enable_lpis (struct vcpu *v) {uint64_t reg = v->domain->arch.vgic.rdist_propbase; -unsigned int nr_lpis =BIT ((reg &0x1f ) +1 ); +unsigned int nr_lpis =BIT ((reg &0x1f ) +1 ,UL );/* rdists_enabled is protected by the domain lock. */ ASSERT (spin_is_locked (&v->domain->arch.vgic.lock));xen/arch/arm/vgic.c +1
-1@@ -65,7 +65,7 @@ struct vgic_irq_rank *vgic_rank_irq(struct vcpu *v, unsigned int irq)void vgic_init_pending_irq (struct pending_irq *p,unsigned int virq) {/* The lpi_vcpu_id field must be big enough to hold a VCPU ID. */ -BUILD_BUG_ON (BIT (sizeof (p->lpi_vcpu_id) *8 ) <MAX_VIRT_CPUS ); +BUILD_BUG_ON (BIT (sizeof (p->lpi_vcpu_id) *8 ,UL ) <MAX_VIRT_CPUS );memset (p,0 ,sizeof (*p));INIT_LIST_HEAD (&p->inflight);xen/drivers/char/meson-uart.c +8
-8@@ -31,16 +31,16 @@#define AML_UART_MISC_REG 0x10/* UART_CONTROL bits */ -#define AML_UART_TX_RST BIT(22) -#define AML_UART_RX_RST BIT(23) -#define AML_UART_CLEAR_ERR BIT(24) -#define AML_UART_RX_INT_EN BIT(27) -#define AML_UART_TX_INT_EN BIT(28) +#define AML_UART_TX_RST BIT(22, UL) +#define AML_UART_RX_RST BIT(23, UL) +#define AML_UART_CLEAR_ERR BIT(24, UL) +#define AML_UART_RX_INT_EN BIT(27, UL) +#define AML_UART_TX_INT_EN BIT(28, UL)/* UART_STATUS bits */ -#define AML_UART_RX_FIFO_EMPTY BIT(20) -#define AML_UART_TX_FIFO_FULL BIT(21) -#define AML_UART_TX_FIFO_EMPTY BIT(22) +#define AML_UART_RX_FIFO_EMPTY BIT(20, UL) +#define AML_UART_TX_FIFO_FULL BIT(21, UL) +#define AML_UART_TX_FIFO_EMPTY BIT(22, UL)#define AML_UART_TX_CNT_MASK GENMASK(14, 8)/* AML_UART_MISC bits */ xen/drivers/char/mvebu-uart.c +17
-17@@ -29,27 +29,27 @@#define UART_TX_REG 0x04#define UART_CTRL_REG 0x08 -#define CTRL_TXFIFO_RST BIT(15) -#define CTRL_RXFIFO_RST BIT(14) -#define CTRL_TX_RDY_INT BIT(5) -#define CTRL_RX_RDY_INT BIT(4) -#define CTRL_BRK_DET_INT BIT(3) -#define CTRL_FRM_ERR_INT BIT(2) -#define CTRL_PAR_ERR_INT BIT(1) -#define CTRL_OVR_ERR_INT BIT(0) +#define CTRL_TXFIFO_RST BIT(15, UL) +#define CTRL_RXFIFO_RST BIT(14, UL) +#define CTRL_TX_RDY_INT BIT(5, UL) +#define CTRL_RX_RDY_INT BIT(4, UL) +#define CTRL_BRK_DET_INT BIT(3, UL) +#define CTRL_FRM_ERR_INT BIT(2, UL) +#define CTRL_PAR_ERR_INT BIT(1, UL) +#define CTRL_OVR_ERR_INT BIT(0, UL)#define CTRL_ERR_INT (CTRL_BRK_DET_INT | CTRL_FRM_ERR_INT | \ CTRL_PAR_ERR_INT | CTRL_OVR_ERR_INT)#define UART_STATUS_REG 0x0c -#define STATUS_TXFIFO_EMP BIT(13) -#define STATUS_TXFIFO_FUL BIT(11) -#define STATUS_TXFIFO_HFL BIT(10) -#define STATUS_TX_RDY BIT(5) -#define STATUS_RX_RDY BIT(4) -#define STATUS_BRK_DET BIT(3) -#define STATUS_FRM_ERR BIT(2) -#define STATUS_PAR_ERR BIT(1) -#define STATUS_OVR_ERR BIT(0) +#define STATUS_TXFIFO_EMP BIT(13, UL) +#define STATUS_TXFIFO_FUL BIT(11, UL) +#define STATUS_TXFIFO_HFL BIT(10, UL) +#define STATUS_TX_RDY BIT(5, UL) +#define STATUS_RX_RDY BIT(4, UL) +#define STATUS_BRK_DET BIT(3, UL) +#define STATUS_FRM_ERR BIT(2, UL) +#define STATUS_PAR_ERR BIT(1, UL) +#define STATUS_OVR_ERR BIT(0, UL)#define STATUS_BRK_ERR (STATUS_BRK_DET | STATUS_FRM_ERR | \ STATUS_PAR_ERR | STATUS_OVR_ERR)xen/include/asm-arm/bitops.h +0
-2@@ -21,10 +21,8 @@#define __clear_bit (n,p) clear_bit(n,p)#define BITS_PER_WORD 32 -#define BIT (nr) (1UL << (nr))#define BIT_MASK (nr) (1UL << ((nr) % BITS_PER_WORD))#define BIT_WORD (nr) ((nr) / BITS_PER_WORD) -#define BIT_ULL (nr) (1ULL << (nr))#define BITS_PER_BYTE 8#define ADDR (*(volatile int *) addr)xen/include/asm-arm/gic_v3_defs.h +2
-2@@ -149,9 +149,9 @@ (7UL << GICR_PENDBASER_INNER_CACHEABILITY_SHIFT)#define GICR_PENDBASER_OUTER_CACHEABILITY_MASK \ (7UL << GICR_PENDBASER_OUTER_CACHEABILITY_SHIFT) -#define GICR_PENDBASER_PTZ BIT(62) +#define GICR_PENDBASER_PTZ BIT(62, UL)#define GICR_PENDBASER_RES0_MASK \ - (BIT(63) | GENMASK(61, 59) | GENMASK(55, 52) | \ + (BIT(63, UL) | GENMASK(61, 59) | GENMASK(55, 52) | \ GENMASK(15, 12) | GENMASK(6, 0))#define DEFAULT_PMR_VALUE 0xffxen/include/asm-arm/gic_v3_its.h +5
-5@@ -38,12 +38,12 @@#define GITS_PIDR2 GICR_PIDR2/* Register bits */ -#define GITS_VALID_BIT BIT(63) +#define GITS_VALID_BIT BIT(63, UL) -#define GITS_CTLR_QUIESCENT BIT(31) -#define GITS_CTLR_ENABLE BIT(0) +#define GITS_CTLR_QUIESCENT BIT(31, UL) +#define GITS_CTLR_ENABLE BIT(0, UL) -#define GITS_TYPER_PTA BIT(19) +#define GITS_TYPER_PTA BIT(19, UL)#define GITS_TYPER_DEVIDS_SHIFT 13#define GITS_TYPER_DEVIDS_MASK (0x1fUL << GITS_TYPER_DEVIDS_SHIFT)#define GITS_TYPER_DEVICE_ID_BITS (r) (((r & GITS_TYPER_DEVIDS_MASK) >> \@@ -60,7 +60,7 @@ GITS_TYPER_ITT_SIZE_SHIFT) + 1)#define GITS_TYPER_PHYSICAL (1U << 0) -#define GITS_BASER_INDIRECT BIT(62) +#define GITS_BASER_INDIRECT BIT(62, UL)#define GITS_BASER_INNER_CACHEABILITY_SHIFT 59#define GITS_BASER_TYPE_SHIFT 56#define GITS_BASER_TYPE_MASK (7ULL << GITS_BASER_TYPE_SHIFT)xen/include/xen/const.h +2
-0@@ -21,4 +21,6 @@#define _AT (T ,X ) ((T)(X))#endif +#define BIT (pos, sfx) (_AC(1, sfx) << (pos)) +#endif /* __XEN_CONST_H__ */ Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Andrii Anisov
Reviewed-byRe: [PATCH MM-PART2 RESEND v2 01/19] xen/const: Extend the existing macro BIT to take a suffix in parameter
Quoted 13 lines · Stefano Stabellini
Quoted 12 lines · Julien Grall
Arm currently provides two macro BIT and
BIT_ULLthat are only usable in C and return respectively unsigned long and unsigned long long.Extending the macros to deal with assembly would be a nice benefits as it could replace the common pattern to define fields (
AC(1, sfx)<< X) easier to read.Rather than extending the two macros, it was decided to drop
BIT_ULL()and extend the macroBIT()to take a suffix (e.g U, UL, ULL) in parameter. This would allow to use different suffix without having to define new macros.
Quoted 5 lines
Quoted 5 lines
The new extend macro is now moved in
include/xen/const.hso it can be used by anyone in Xen and also avoid to includebitops.hin assembly code.Signed-off-by Julien Grall <[email protected]>
Signature · 7 lines
Sincerely, Andrii Anisov. _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 2/19 xen/arm: Rename SCTLR_* defines and remove unused one
In reply to Julien GrallThe SCTLR_* are currently used for SCTLR/HSCTLR (arm32) and SCTLR_EL1/SCTLR_EL2 (arm64).
The naming scheme is actually quite confusing because they may only be defined for an archicture (or even an exception level). So it is not easy for the developer to know which one to use.
The naming scheme is reworked by adding Axx_ELx in each define:
* xx is replaced by 32 or 64 if specific to an architecture
* x is replaced by 2 (hypervisor) or 1 (kernel) if specific to an exception level
While doing the renaming, remove the unused defines (or at least the ones that are unlikely going to be used).
---
Changes in v2:
- Fix build on arm32
- Add Andrii's reviewed-by
--- xen/arch/arm/arm32/head.S | 5 +++-- xen/arch/arm/arm64/head.S | 4 ++-- xen/arch/arm/guest_walk.c | 2 +- xen/arch/arm/mm.c | 2 +- xen/arch/arm/traps.c | 6 +++--- xen/include/asm-arm/p2m.h | 4 +++- xen/include/asm-arm/processor.h | 37 +++++++++++++++++-------------------- 7 files changed, 30 insertions(+), 30 deletions(-)
xen/arch/arm/arm32/head.S +3 -2
@@ -244,7 +244,7 @@ cpu_init_done: * Alignment checking enabled, * MMU translation disabled (for now). */ - ldr r0, =(HSCTLR_BASE|SCTLR_A) + ldr r0, =(HSCTLR_BASE|SCTLR_Axx_ELx_A) mcr CP32(r0, HSCTLR) /*
@@ -369,7 +369,8 @@ virtphys_clash: ldr r1, =paging /* Explicit vaddr, not RIP-relative */ mrc CP32(r0, HSCTLR) - orr r0, r0, #(SCTLR_M|SCTLR_C) /* Enable MMU and D-cache */ + /* Enable MMU and D-cache */ + orr r0, r0, #(SCTLR_Axx_ELx_M|SCTLR_Axx_ELx_C) dsb /* Flush PTE writes and finish reads */ mcr CP32(r0, HSCTLR) /* now paging is enabled */ isb /* Now, flush the icache */
xen/arch/arm/arm64/head.S +2 -2
@@ -514,8 +514,8 @@ virtphys_clash: ldr x1, =paging /* Explicit vaddr, not RIP-relative */ mrs x0, SCTLR_EL2 - orr x0, x0, #SCTLR_M /* Enable MMU */ - orr x0, x0, #SCTLR_C /* Enable D-cache */ + orr x0, x0, #SCTLR_Axx_ELx_M /* Enable MMU */ + orr x0, x0, #SCTLR_Axx_ELx_C /* Enable D-cache */ dsb sy /* Flush PTE writes and finish reads */ msr SCTLR_EL2, x0 /* now paging is enabled */ isb /* Now, flush the icache */
xen/arch/arm/guest_walk.c +1 -1
@@ -612,7 +612,7 @@ bool guest_walk_tables(const struct vcpu *v, vaddr_t gva, *perms =GV2M_READ ;/* If the MMU is disabled, there is no need to translate the gva. */ -if ( !(sctlr &SCTLR_M ) ) +if ( !(sctlr & SCTLR_Axx_ELx_M) ) { *ipa = gva;
xen/arch/arm/mm.c +1 -1
@@ -609,7 +609,7 @@ void __init remove_early_mappings(void)*/ static void xen_pt_enforce_wnx (void ) { -WRITE_SYSREG32 (READ_SYSREG32 (SCTLR_EL2 ) |SCTLR_WXN ,SCTLR_EL2 ); +WRITE_SYSREG32 (READ_SYSREG32 (SCTLR_EL2 ) | SCTLR_Axx_ELx_WXN,SCTLR_EL2 );/* * The TLBs may cache SCTLR_EL2.WXN. So ensure it is synchronized * before flushing the TLBs.
xen/arch/arm/traps.c +3 -3
@@ -392,9 +392,9 @@ static void cpsr_switch_mode(struct cpu_user_regs *regs, int mode) regs->cpsr |=PSR_IRQ_MASK ;if ( mode ==PSR_MODE_ABT ) regs->cpsr |=PSR_ABT_MASK ; -if ( sctlr &SCTLR_TE ) +if ( sctlr & SCTLR_A32_ELx_TE ) regs->cpsr |=PSR_THUMB ; -if ( sctlr &SCTLR_EE ) +if ( sctlr & SCTLR_Axx_ELx_EE ) regs->cpsr |=PSR_BIG_ENDIAN ; }
@@ -402,7 +402,7 @@ static vaddr_t exception_handler32(vaddr_t offset) {uint32_t sctlr =READ_SYSREG32 (SCTLR_EL1 ); -if ( sctlr &SCTLR_V ) +if ( sctlr &SCTLR_A32_EL1_V )return 0xffff0000 + offset;else /* always have security exceptions */ return READ_SYSREG (VBAR_EL1 ) + offset;
xen/include/asm-arm/p2m.h +3 -1
@@ -391,10 +391,12 @@ static inline int set_foreign_p2m_entry(struct domain *d, unsigned long gfn,*/ static inline bool vcpu_has_cache_enabled (struct vcpu *v) { +const uint32_t mask = SCTLR_Axx_ELx_C | SCTLR_Axx_ELx_M; +/* Only works with the current vCPU */ ASSERT (current == v); -return (READ_SYSREG32 (SCTLR_EL1 ) & (SCTLR_C |SCTLR_M )) == (SCTLR_C |SCTLR_M ); +return (READ_SYSREG32 (SCTLR_EL1 ) & mask) == mask; }#endif /* _XEN_P2M_H */
xen/include/asm-arm/processor.h +17 -20
@@ -117,26 +117,23 @@#define TTBCR_PD1 (_AC(1,U)<<5)/* SCTLR System Control Register. */ -/* HSCTLR is a subset of this. */ -#define SCTLR_TE (_AC(1,U)<<30) -#define SCTLR_AFE (_AC(1,U)<<29) -#define SCTLR_TRE (_AC(1,U)<<28) -#define SCTLR_NMFI (_AC(1,U)<<27) -#define SCTLR_EE (_AC(1,U)<<25) -#define SCTLR_VE (_AC(1,U)<<24) -#define SCTLR_U (_AC(1,U)<<22) -#define SCTLR_FI (_AC(1,U)<<21) -#define SCTLR_WXN (_AC(1,U)<<19) -#define SCTLR_HA (_AC(1,U)<<17) -#define SCTLR_RR (_AC(1,U)<<14) -#define SCTLR_V (_AC(1,U)<<13) -#define SCTLR_I (_AC(1,U)<<12) -#define SCTLR_Z (_AC(1,U)<<11) -#define SCTLR_SW (_AC(1,U)<<10) -#define SCTLR_B (_AC(1,U)<<7) -#define SCTLR_C (_AC(1,U)<<2) -#define SCTLR_A (_AC(1,U)<<1) -#define SCTLR_M (_AC(1,U)<<0) + +/* Bits specific to SCTLR_EL1 for Arm32 */ + +#define SCTLR_A32_EL1_V (_AC(1,U)<<13) + +/* Common bits for SCTLR_ELx for Arm32 */ + +#define SCTLR_A32_ELx_TE (_AC(1,U)<<30) +#define SCTLR_A32_ELx_FI (_AC(1,U)<<21) + +/* Common bits for SCTLR_ELx on all architectures */ +#define SCTLR_Axx_ELx_EE (_AC(1,U)<<25) +#define SCTLR_Axx_ELx_WXN (_AC(1,U)<<19) +#define SCTLR_Axx_ELx_I (_AC(1,U)<<12) +#define SCTLR_Axx_ELx_C (_AC(1,U)<<2) +#define SCTLR_Axx_ELx_A (_AC(1,U)<<1) +#define SCTLR_Axx_ELx_M (_AC(1,U)<<0)#define HSCTLR_BASE _AC(0x30c51878,U)
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 02/19] xen/arm: Rename SCTLR_* defines and remove unused one
Quoted 12 lines · Julien Grall
The SCTLR_* are currently used for SCTLR/HSCTLR (arm32) and SCTLR_EL1/SCTLR_EL2 (arm64).
The naming scheme is actually quite confusing because they may only be defined for an archicture (or even an exception level). So it is not easy for the developer to know which one to use.
The naming scheme is reworked by adding
Axx_ELxin each define:
* xx is replaced by 32 or 64 if specific to an architecture
* x is replaced by 2 (hypervisor) or 1 (kernel) if specific to an exception level
Quoted 5 lines
While doing the renaming, remove the unused defines (or at least the ones that are unlikely going to be used).
Signed-off-by Julien Grall <[email protected]>Reviewed-by Andrii Anisov <[email protected]>
Quoted 172 lines
---
Changes in v2:
- Fix build on arm32
- Add Andrii's reviewed-by--- xen/arch/arm/arm32/head.S | 5 +++-- xen/arch/arm/arm64/head.S | 4 ++-- xen/arch/arm/guest_walk.c | 2 +- xen/arch/arm/mm.c | 2 +- xen/arch/arm/traps.c | 6 +++--- xen/include/asm-arm/p2m.h | 4 +++- xen/include/asm-arm/processor.h | 37 +++++++++++++++++-------------------- 7 files changed, 30 insertions(+), 30 deletions(-)xen/arch/arm/arm32/head.S +3
-2@@ -244,7 +244,7 @@ cpu_init_done: * Alignment checking enabled, * MMU translation disabled (for now). */ - ldr r0, =(HSCTLR_BASE|SCTLR_A) + ldr r0, =(HSCTLR_BASE|SCTLR_Axx_ELx_A) mcr CP32(r0, HSCTLR) /*@@ -369,7 +369,8 @@ virtphys_clash: ldr r1, =paging /* Explicit vaddr, not RIP-relative */ mrc CP32(r0, HSCTLR) - orr r0, r0, #(SCTLR_M|SCTLR_C) /* Enable MMU and D-cache */ + /* Enable MMU and D-cache */ + orr r0, r0, #(SCTLR_Axx_ELx_M|SCTLR_Axx_ELx_C) dsb /* Flush PTE writes and finish reads */ mcr CP32(r0, HSCTLR) /* now paging is enabled */ isb /* Now, flush the icache */xen/arch/arm/arm64/head.S +2
-2@@ -514,8 +514,8 @@ virtphys_clash: ldr x1, =paging /* Explicit vaddr, not RIP-relative */ mrs x0, SCTLR_EL2 - orr x0, x0, #SCTLR_M /* Enable MMU */ - orr x0, x0, #SCTLR_C /* Enable D-cache */ + orr x0, x0, #SCTLR_Axx_ELx_M /* Enable MMU */ + orr x0, x0, #SCTLR_Axx_ELx_C /* Enable D-cache */ dsb sy /* Flush PTE writes and finish reads */ msr SCTLR_EL2, x0 /* now paging is enabled */ isb /* Now, flush the icache */xen/arch/arm/guest_walk.c +1
-1@@ -612,7 +612,7 @@ bool guest_walk_tables(const struct vcpu *v, vaddr_t gva, *perms =GV2M_READ ;/* If the MMU is disabled, there is no need to translate the gva. */ -if ( !(sctlr &SCTLR_M ) ) +if ( !(sctlr & SCTLR_Axx_ELx_M) ) { *ipa = gva;xen/arch/arm/mm.c +1
-1@@ -609,7 +609,7 @@ void __init remove_early_mappings(void)*/ static void xen_pt_enforce_wnx (void ) { -WRITE_SYSREG32 (READ_SYSREG32 (SCTLR_EL2 ) |SCTLR_WXN ,SCTLR_EL2 ); +WRITE_SYSREG32 (READ_SYSREG32 (SCTLR_EL2 ) | SCTLR_Axx_ELx_WXN,SCTLR_EL2 );/* * The TLBs may cache SCTLR_EL2.WXN. So ensure it is synchronized * before flushing the TLBs. xen/arch/arm/traps.c +3
-3@@ -392,9 +392,9 @@ static void cpsr_switch_mode(struct cpu_user_regs *regs, int mode) regs->cpsr |=PSR_IRQ_MASK ;if ( mode ==PSR_MODE_ABT ) regs->cpsr |=PSR_ABT_MASK ; -if ( sctlr &SCTLR_TE ) +if ( sctlr & SCTLR_A32_ELx_TE ) regs->cpsr |=PSR_THUMB ; -if ( sctlr &SCTLR_EE ) +if ( sctlr & SCTLR_Axx_ELx_EE ) regs->cpsr |=PSR_BIG_ENDIAN ; }@@ -402,7 +402,7 @@ static vaddr_t exception_handler32(vaddr_t offset) {uint32_t sctlr =READ_SYSREG32 (SCTLR_EL1 ); -if ( sctlr &SCTLR_V ) +if ( sctlr &SCTLR_A32_EL1_V )return 0xffff0000 + offset;else /* always have security exceptions */ return READ_SYSREG (VBAR_EL1 ) + offset;xen/include/asm-arm/p2m.h +3
-1@@ -391,10 +391,12 @@ static inline int set_foreign_p2m_entry(struct domain *d, unsigned long gfn,*/ static inline bool vcpu_has_cache_enabled (struct vcpu *v) { +const uint32_t mask = SCTLR_Axx_ELx_C | SCTLR_Axx_ELx_M; +/* Only works with the current vCPU */ ASSERT (current == v); -return (READ_SYSREG32 (SCTLR_EL1 ) & (SCTLR_C |SCTLR_M )) == (SCTLR_C |SCTLR_M ); +return (READ_SYSREG32 (SCTLR_EL1 ) & mask) == mask; }#endif /* _XEN_P2M_H */ xen/include/asm-arm/processor.h +17
-20@@ -117,26 +117,23 @@#define TTBCR_PD1 (_AC(1,U)<<5)/* SCTLR System Control Register. */ -/* HSCTLR is a subset of this. */ -#define SCTLR_TE (_AC(1,U)<<30) -#define SCTLR_AFE (_AC(1,U)<<29) -#define SCTLR_TRE (_AC(1,U)<<28) -#define SCTLR_NMFI (_AC(1,U)<<27) -#define SCTLR_EE (_AC(1,U)<<25) -#define SCTLR_VE (_AC(1,U)<<24) -#define SCTLR_U (_AC(1,U)<<22) -#define SCTLR_FI (_AC(1,U)<<21) -#define SCTLR_WXN (_AC(1,U)<<19) -#define SCTLR_HA (_AC(1,U)<<17) -#define SCTLR_RR (_AC(1,U)<<14) -#define SCTLR_V (_AC(1,U)<<13) -#define SCTLR_I (_AC(1,U)<<12) -#define SCTLR_Z (_AC(1,U)<<11) -#define SCTLR_SW (_AC(1,U)<<10) -#define SCTLR_B (_AC(1,U)<<7) -#define SCTLR_C (_AC(1,U)<<2) -#define SCTLR_A (_AC(1,U)<<1) -#define SCTLR_M (_AC(1,U)<<0) + +/* Bits specific to SCTLR_EL1 for Arm32 */ + +#define SCTLR_A32_EL1_V (_AC(1,U)<<13) + +/* Common bits for SCTLR_ELx for Arm32 */ + +#define SCTLR_A32_ELx_TE (_AC(1,U)<<30) +#define SCTLR_A32_ELx_FI (_AC(1,U)<<21) + +/* Common bits for SCTLR_ELx on all architectures */ +#define SCTLR_Axx_ELx_EE (_AC(1,U)<<25) +#define SCTLR_Axx_ELx_WXN (_AC(1,U)<<19) +#define SCTLR_Axx_ELx_I (_AC(1,U)<<12) +#define SCTLR_Axx_ELx_C (_AC(1,U)<<2) +#define SCTLR_Axx_ELx_A (_AC(1,U)<<1) +#define SCTLR_Axx_ELx_M (_AC(1,U)<<0)#define HSCTLR_BASE _AC(0x30c51878,U)Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 3/19 xen/arm: processor: Use BIT(.., UL) instead of _AC(1, U) in SCTLR_ defines
In reply to Julien GrallUse the pattern BIT(..., UL) to make the code more readable. Note that unsigned long is used instead of unsigned because SCTLR is technically 32-bit on Arm32 and 64-bit on Arm64.
---
Changes in v2:
- Rework the patch to use BIT(..., UL) instead of _BITUL(...).
--- xen/include/asm-arm/processor.h | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-)
xen/include/asm-arm/processor.h +9 -9
@@ -120,20 +120,20 @@/* Bits specific to SCTLR_EL1 for Arm32 */ -#define SCTLR_A32_EL1_V (_AC(1,U)<<13) +#define SCTLR_A32_EL1_V BIT(13, UL)/* Common bits for SCTLR_ELx for Arm32 */ -#define SCTLR_A32_ELx_TE (_AC(1,U)<<30) -#define SCTLR_A32_ELx_FI (_AC(1,U)<<21) +#define SCTLR_A32_ELx_TE BIT(30, UL) +#define SCTLR_A32_ELx_FI BIT(21, UL)/* Common bits for SCTLR_ELx on all architectures */ -#define SCTLR_Axx_ELx_EE (_AC(1,U)<<25) -#define SCTLR_Axx_ELx_WXN (_AC(1,U)<<19) -#define SCTLR_Axx_ELx_I (_AC(1,U)<<12) -#define SCTLR_Axx_ELx_C (_AC(1,U)<<2) -#define SCTLR_Axx_ELx_A (_AC(1,U)<<1) -#define SCTLR_Axx_ELx_M (_AC(1,U)<<0) +#define SCTLR_Axx_ELx_EE BIT(25, UL) +#define SCTLR_Axx_ELx_WXN BIT(19, UL) +#define SCTLR_Axx_ELx_I BIT(12, UL) +#define SCTLR_Axx_ELx_C BIT(2, UL) +#define SCTLR_Axx_ELx_A BIT(1, UL) +#define SCTLR_Axx_ELx_M BIT(0, UL)#define HSCTLR_BASE _AC(0x30c51878,U)
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 03/19] xen/arm: processor: Use BIT(.., UL) instead of _AC(1, U) in SCTLR_ defines
Quoted 5 lines · Julien Grall
Use the pattern
BIT(..., UL)to make the code more readable. Note that unsigned long is used instead of unsigned because SCTLR is technically 32-bit on Arm32 and 64-bit on Arm64.Signed-off-by Julien Grall <[email protected]>
Quoted 44 lines
---
Changes in v2:
- Rework the patch to useBIT(..., UL)instead of_BITUL(...).--- xen/include/asm-arm/processor.h | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-)xen/include/asm-arm/processor.h +9
-9@@ -120,20 +120,20 @@/* Bits specific to SCTLR_EL1 for Arm32 */ -#define SCTLR_A32_EL1_V (_AC(1,U)<<13) +#define SCTLR_A32_EL1_V BIT(13, UL)/* Common bits for SCTLR_ELx for Arm32 */ -#define SCTLR_A32_ELx_TE (_AC(1,U)<<30) -#define SCTLR_A32_ELx_FI (_AC(1,U)<<21) +#define SCTLR_A32_ELx_TE BIT(30, UL) +#define SCTLR_A32_ELx_FI BIT(21, UL)/* Common bits for SCTLR_ELx on all architectures */ -#define SCTLR_Axx_ELx_EE (_AC(1,U)<<25) -#define SCTLR_Axx_ELx_WXN (_AC(1,U)<<19) -#define SCTLR_Axx_ELx_I (_AC(1,U)<<12) -#define SCTLR_Axx_ELx_C (_AC(1,U)<<2) -#define SCTLR_Axx_ELx_A (_AC(1,U)<<1) -#define SCTLR_Axx_ELx_M (_AC(1,U)<<0) +#define SCTLR_Axx_ELx_EE BIT(25, UL) +#define SCTLR_Axx_ELx_WXN BIT(19, UL) +#define SCTLR_Axx_ELx_I BIT(12, UL) +#define SCTLR_Axx_ELx_C BIT(2, UL) +#define SCTLR_Axx_ELx_A BIT(1, UL) +#define SCTLR_Axx_ELx_M BIT(0, UL)#define HSCTLR_BASE _AC(0x30c51878,U)Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 03/19] xen/arm: processor: Use BIT(.., UL) instead of _AC(1, U) in SCTLR_ defines
In reply to Julien GrallQuoted 5 lines · Julien Grall
Use the pattern
BIT(..., UL)to make the code more readable. Note that unsigned long is used instead of unsigned because SCTLR is technically 32-bit on Arm32 and 64-bit on Arm64.Signed-off-by Julien Grall <[email protected]>
Signature · 7 lines
Sincerely, Andrii Anisov. _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 4/19 xen/arm: Rework HSCTLR_BASE
In reply to Julien GrallThe current value of HSCTLR_BASE for Arm64 is pretty wrong. It would actually turn on SCTLR_EL2.nAA (bit 6) on hardware implementing ARMv8.4-LSE.
Furthermore, the documentation of what is cleared/set in SCTLR_EL2 is also not correct and looks like to be a verbatim copy from Arm32.
HSCTLR_BASE is replaced with a bunch of per-architecture new defines helping to understand better what is the initialie value for SCTLR_EL2/HSCTLR.
Note the defines *_CLEAR are only used to check the state of each bits are known.
Lastly, the documentation is dropped from arm{32,64}/head.S as it would be pretty easy to get out-of-sync with the definitions.
---
Changes in v2:
- Use BIT(..., UL) instead of _BITUL
--- xen/arch/arm/arm32/head.S | 12 +-------- xen/arch/arm/arm64/head.S | 10 +------- xen/include/asm-arm/processor.h | 54 ++++++++++++++++++++++++++++++++++++++++- 3 files changed, 55 insertions(+), 21 deletions(-)
xen/arch/arm/arm32/head.S +1 -11
@@ -234,17 +234,7 @@ cpu_init_done: ldr r0, =(TCR_RES1|TCR_SH0_IS|TCR_ORGN0_WBWA|TCR_IRGN0_WBWA|TCR_T0SZ(0)) mcr CP32(r0, HTCR) - /* - * Set up the HSCTLR: - * Exceptions in LE ARM, - * Low-latency IRQs disabled, - * Write-implies-XN disabled (for now), - * D-cache disabled (for now), - * I-cache enabled, - * Alignment checking enabled, - * MMU translation disabled (for now). - */ - ldr r0, =(HSCTLR_BASE|SCTLR_Axx_ELx_A) + ldr r0, =HSCTLR_SET mcr CP32(r0, HSCTLR) /*
xen/arch/arm/arm64/head.S +1 -9
@@ -363,15 +363,7 @@ skip_bss: msr tcr_el2, x0 - /* Set up the SCTLR_EL2: - * Exceptions in LE ARM, - * Low-latency IRQs disabled, - * Write-implies-XN disabled (for now), - * D-cache disabled (for now), - * I-cache enabled, - * Alignment checking disabled, - * MMU translation disabled (for now). */ - ldr x0, =(HSCTLR_BASE) + ldr x0, =SCTLR_EL2_SET msr SCTLR_EL2, x0 /* Ensure that any exceptions encountered at EL2
xen/include/asm-arm/processor.h +53 -1
66 lines not on this page · raw message
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 04/19] xen/arm: Rework HSCTLR_BASE
Quoted 7 lines · Julien Grall
The current value of
HSCTLR_BASEfor Arm64 is pretty wrong. It would actually turn on SCTLR_EL2.nAA (bit 6) on hardware implementing ARMv8.4-LSE.Furthermore, the documentation of what is cleared/set in
SCTLR_EL2is also not correct and looks like to be a verbatim copy from Arm32.
Quoted 6 lines
HSCTLR_BASEis replaced with a bunch of per-architecture new defines helping to understand better what is the initialie value for SCTLR_EL2/HSCTLR.Note the defines
*_CLEARare only used to check the state of each bits are known.
So basically the only purpose of HSCTLR_CLEAR is to execute:
#if (HSCTLR_SET ^ HSCTLR_CLEAR) != 0xffffffffU
Right? It is good to have the check.
Please add a one-line comment on top of HSCTLR_CLEAR -- "only used to check that the state of each bits are known".
Quoted 81 lines
Lastly, the documentation is dropped from
arm{32,64}/head.Sas it would be pretty easy to get out-of-sync with the definitions.Signed-off-by Julien Grall <[email protected]>---
Changes in v2:
- UseBIT(..., UL)instead of_BITUL--- xen/arch/arm/arm32/head.S | 12 +-------- xen/arch/arm/arm64/head.S | 10 +------- xen/include/asm-arm/processor.h | 54 ++++++++++++++++++++++++++++++++++++++++- 3 files changed, 55 insertions(+), 21 deletions(-)xen/arch/arm/arm32/head.S +1
-11@@ -234,17 +234,7 @@ cpu_init_done: ldr r0, =(TCR_RES1|TCR_SH0_IS|TCR_ORGN0_WBWA|TCR_IRGN0_WBWA|TCR_T0SZ(0)) mcr CP32(r0, HTCR) - /* - * Set up the HSCTLR: - * Exceptions in LE ARM, - * Low-latency IRQs disabled, - * Write-implies-XN disabled (for now), - * D-cache disabled (for now), - * I-cache enabled, - * Alignment checking enabled, - * MMU translation disabled (for now). - */ - ldr r0, =(HSCTLR_BASE|SCTLR_Axx_ELx_A) + ldr r0, =HSCTLR_SET mcr CP32(r0, HSCTLR) /*xen/arch/arm/arm64/head.S +1
-9@@ -363,15 +363,7 @@ skip_bss: msr tcr_el2, x0 - /* Set up the SCTLR_EL2: - * Exceptions in LE ARM, - * Low-latency IRQs disabled, - * Write-implies-XN disabled (for now), - * D-cache disabled (for now), - * I-cache enabled, - * Alignment checking disabled, - * MMU translation disabled (for now). */ - ldr x0, =(HSCTLR_BASE) + ldr x0, =SCTLR_EL2_SET msr SCTLR_EL2, x0 /* Ensure that any exceptions encountered at EL2xen/include/asm-arm/processor.h +7
-1@@ -127,6 +127,9 @@#define SCTLR_A32_ELx_TE BIT(30, UL)#define SCTLR_A32_ELx_FI BIT(21, UL) +/* Common bits for SCTLR_ELx for Arm64 */ +#define SCTLR_A64_ELx_SA BIT(3, UL) +/* Common bits for SCTLR_ELx on all architectures */ #define SCTLR_Axx_ELx_EE BIT(25, UL)#define SCTLR_Axx_ELx_WXN BIT(19, UL)@@ -135,7 +138,56 @@#define SCTLR_Axx_ELx_A BIT(1, UL)#define SCTLR_Axx_ELx_M BIT(0, UL) -#define HSCTLR_BASE _AC(0x30c51878,U) +#ifdef CONFIG_ARM_32 + +#define HSCTLR_RES1 (BIT( 3, UL) | BIT( 4, UL) | BIT( 5, UL) |\ + BIT( 6, UL) | BIT(11, UL) | BIT(16, UL) |\
xen/include/asm-arm/processor.h line 154
Quoted 10 lines
xen/include/asm-arm/processor.h +10
-0+ BIT(18, UL) | BIT(22, UL) | BIT(23, UL) |\ + BIT(28, UL) | BIT(29, UL)) + +#define HSCTLR_RES0 (BIT(7, UL) | BIT(8, UL) | BIT(9, UL) | BIT(10, UL) |\ + BIT(13, UL) | BIT(14, UL) | BIT(15, UL) | BIT(17, UL) |\ + BIT(20, UL) | BIT(24, UL) | BIT(26, UL) | BIT(27, UL) |\ + BIT(31, UL)) + +/* Initial value for HSCTLR */ +#define HSCTLR_SET (HSCTLR_RES1 | SCTLR_Axx_ELx_A | SCTLR_Axx_ELx_I)
As far as my calculations go, it looks like the only difference is SCTLR_Axx_ELx_A compared to 0x30c51878, right? Which is the alignment check.
Quoted 15 lines
xen/include/asm-arm/processor.h +15
-0+#define HSCTLR_CLEAR (HSCTLR_RES0 | SCTLR_Axx_ELx_M |\ + SCTLR_Axx_ELx_C | SCTLR_Axx_ELx_WXN |\ + SCTLR_A32_ELx_FI | SCTLR_Axx_ELx_EE |\ + SCTLR_A32_ELx_TE) + +#if (HSCTLR_SET ^HSCTLR_CLEAR ) !=0xffffffffU +#error "Inconsistent HSCTLR set/clear bits" +#endif + +#else + +#define SCTLR_EL2_RES1 (BIT( 4, UL) | BIT( 5, UL) | BIT(11, UL) |\ + BIT(16, UL) | BIT(18, UL) | BIT(22, UL) |\ + BIT(23, UL) | BIT(28, UL) | BIT(29, UL)) +
xen/include/asm-arm/processor.h line 180
Quoted 10 lines
xen/include/asm-arm/processor.h +10
-0+#define SCTLR_EL2_RES0 (BIT( 6, UL) | BIT( 7, UL) | BIT( 8, UL) |\ + BIT( 9, UL) | BIT(10, UL) | BIT(13, UL) |\ + BIT(14, UL) | BIT(15, UL) | BIT(17, UL) |\ + BIT(20, UL) | BIT(21, UL) | BIT(24, UL) |\ + BIT(26, UL) | BIT(27, UL) | BIT(30, UL) |\ + BIT(31, UL) | (0xffffffffULL << 32)) + +/* Initial value for SCTLR_EL2 */ +#define SCTLR_EL2_SET (SCTLR_EL2_RES1 | SCTLR_A64_ELx_SA |\ + SCTLR_Axx_ELx_I)
Same here, you removed the reserved bits, and added the alignment check, same as for aarch32. If I got it right, it would be nice to add a statement like this to the commit message.
xen/include/asm-arm/processor.h line 193
Quoted 12 lines
xen/include/asm-arm/processor.h +9
-0+#define SCTLR_EL2_CLEAR (SCTLR_EL2_RES0 | SCTLR_Axx_ELx_M |\ + SCTLR_Axx_ELx_A | SCTLR_Axx_ELx_C |\ + SCTLR_Axx_ELx_WXN | SCTLR_Axx_ELx_EE) + +#if (SCTLR_EL2_SET ^SCTLR_EL2_CLEAR ) !=0xffffffffffffffffUL +#error "Inconsistent SCTLR_EL2 set/clear bits" +#endif + +#endif /* HCR Hyp Configuration Register */ #define HCR_RW (_AC(1,UL)<<31)/* Register Width, ARM64 only */
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 04/19] xen/arm: Rework HSCTLR_BASE
Hi,
Quoted 10 lines · Stefano Stabellini
Quoted 9 lines · Julien Grall
The current value of
HSCTLR_BASEfor Arm64 is pretty wrong. It would actually turn on SCTLR_EL2.nAA (bit 6) on hardware implementing ARMv8.4-LSE.Furthermore, the documentation of what is cleared/set in
SCTLR_EL2is also not correct and looks like to be a verbatim copy from Arm32.
HSCTLR_BASEis replaced with a bunch of per-architecture new defines helping to understand better what is the initialie value for
s/initialie/initial/
Quoted 5 lines
Quoted 4 lines
SCTLR_EL2/HSCTLR.
Note the defines
*_CLEARare only used to check the state of each bits are known.
Quoted 8 lines
So basically the only purpose of
HSCTLR_CLEARis to execute:#if (HSCTLR_SET ^ HSCTLR_CLEAR) != 0xffffffffURight? It is good to have the check.
Please add a one-line comment on top of
HSCTLR_CLEAR-- "only used to check that the state of each bits are known".
We don't commonly add a comment every time a define is used only one time. So what's the benefits here?
In all honesty, such wording in the commit message was probably over the top. I am thinking to replace the sentence with:
"Lastly, all the bits are now described as either set or cleared. This allows us to check at pre-processing time the consistency of the initial value."
Quoted 89 lines
Quoted 89 lines
Lastly, the documentation is dropped from
arm{32,64}/head.Sas it would be pretty easy to get out-of-sync with the definitions.Signed-off-by Julien Grall <[email protected]>---
Changes in v2:
- UseBIT(..., UL)instead of_BITUL--- xen/arch/arm/arm32/head.S | 12 +-------- xen/arch/arm/arm64/head.S | 10 +------- xen/include/asm-arm/processor.h | 54 ++++++++++++++++++++++++++++++++++++++++- 3 files changed, 55 insertions(+), 21 deletions(-)xen/arch/arm/arm32/head.S +1
-11@@ -234,17 +234,7 @@ cpu_init_done: ldr r0, =(TCR_RES1|TCR_SH0_IS|TCR_ORGN0_WBWA|TCR_IRGN0_WBWA|TCR_T0SZ(0)) mcr CP32(r0, HTCR) - /* - * Set up the HSCTLR: - * Exceptions in LE ARM, - * Low-latency IRQs disabled, - * Write-implies-XN disabled (for now), - * D-cache disabled (for now), - * I-cache enabled, - * Alignment checking enabled, - * MMU translation disabled (for now). - */ - ldr r0, =(HSCTLR_BASE|SCTLR_Axx_ELx_A) + ldr r0, =HSCTLR_SET mcr CP32(r0, HSCTLR) /*xen/arch/arm/arm64/head.S +1
-9@@ -363,15 +363,7 @@ skip_bss: msr tcr_el2, x0 - /* Set up the SCTLR_EL2: - * Exceptions in LE ARM, - * Low-latency IRQs disabled, - * Write-implies-XN disabled (for now), - * D-cache disabled (for now), - * I-cache enabled, - * Alignment checking disabled, - * MMU translation disabled (for now). */ - ldr x0, =(HSCTLR_BASE) + ldr x0, =SCTLR_EL2_SET msr SCTLR_EL2, x0 /* Ensure that any exceptions encountered at EL2xen/include/asm-arm/processor.h +13
-1@@ -127,6 +127,9 @@#define SCTLR_A32_ELx_TE BIT(30, UL)#define SCTLR_A32_ELx_FI BIT(21, UL) +/* Common bits for SCTLR_ELx for Arm64 */ +#define SCTLR_A64_ELx_SA BIT(3, UL) +/* Common bits for SCTLR_ELx on all architectures */ #define SCTLR_Axx_ELx_EE BIT(25, UL)#define SCTLR_Axx_ELx_WXN BIT(19, UL)@@ -135,7 +138,56 @@#define SCTLR_Axx_ELx_A BIT(1, UL)#define SCTLR_Axx_ELx_M BIT(0, UL) -#define HSCTLR_BASE _AC(0x30c51878,U) +#ifdef CONFIG_ARM_32 + +#define HSCTLR_RES1 (BIT( 3, UL) | BIT( 4, UL) | BIT( 5, UL) |\ + BIT( 6, UL) | BIT(11, UL) | BIT(16, UL) |\ + BIT(18, UL) | BIT(22, UL) | BIT(23, UL) |\ + BIT(28, UL) | BIT(29, UL)) + +#define HSCTLR_RES0 (BIT(7, UL) | BIT(8, UL) | BIT(9, UL) | BIT(10, UL) |\ + BIT(13, UL) | BIT(14, UL) | BIT(15, UL) | BIT(17, UL) |\ + BIT(20, UL) | BIT(24, UL) | BIT(26, UL) | BIT(27, UL) |\
Quoted 8 lines
xen/include/asm-arm/processor.h line 154
Quoted 4 lines
xen/include/asm-arm/processor.h +4
-0+ BIT(31, UL)) + +/* Initial value for HSCTLR */ +#define HSCTLR_SET (HSCTLR_RES1 | SCTLR_Axx_ELx_A | SCTLR_Axx_ELx_I)As far as my calculations go, it looks like the only difference is
SCTLR_Axx_ELx_Acompared to0x30c51878, right? Which is the alignment check.
That's correct and match the initial value on Arm32 (see HSCTR_SET | SCTLR_Axx_ELx_A in the head.S).
Quoted 23 lines
Quoted 23 lines
xen/include/asm-arm/processor.h +21
-0+#define HSCTLR_CLEAR (HSCTLR_RES0 | SCTLR_Axx_ELx_M |\ + SCTLR_Axx_ELx_C | SCTLR_Axx_ELx_WXN |\ + SCTLR_A32_ELx_FI | SCTLR_Axx_ELx_EE |\ + SCTLR_A32_ELx_TE) + +#if (HSCTLR_SET ^HSCTLR_CLEAR ) !=0xffffffffU +#error "Inconsistent HSCTLR set/clear bits" +#endif + +#else + +#define SCTLR_EL2_RES1 (BIT( 4, UL) | BIT( 5, UL) | BIT(11, UL) |\ + BIT(16, UL) | BIT(18, UL) | BIT(22, UL) |\ + BIT(23, UL) | BIT(28, UL) | BIT(29, UL)) + +#define SCTLR_EL2_RES0 (BIT( 6, UL) | BIT( 7, UL) | BIT( 8, UL) |\ + BIT( 9, UL) | BIT(10, UL) | BIT(13, UL) |\ + BIT(14, UL) | BIT(15, UL) | BIT(17, UL) |\ + BIT(20, UL) | BIT(21, UL) | BIT(24, UL) |\ + BIT(26, UL) | BIT(27, UL) | BIT(30, UL) |\ + BIT(31, UL) | (0xffffffffULL << 32))
Quoted 8 lines
xen/include/asm-arm/processor.h line 180
Quoted 4 lines
xen/include/asm-arm/processor.h +4
-0+ +/* Initial value for SCTLR_EL2 */ +#define SCTLR_EL2_SET (SCTLR_EL2_RES1 | SCTLR_A64_ELx_SA |\ + SCTLR_Axx_ELx_I)Same here, you removed the reserved bits, and added the alignment check, same as for aarch32. If I got it right, it would be nice to add a statement like this to the commit message.
I don't see why "reserved bits" I dropped nor which alignment check I added.
It would be extremely useful if you provide more details in your review... In this case, it would be the exact bits I dropped/added.
Quoted 12 lines
Quoted 12 lines
xen/include/asm-arm/processor.h +9
-0+#define SCTLR_EL2_CLEAR (SCTLR_EL2_RES0 | SCTLR_Axx_ELx_M |\ + SCTLR_Axx_ELx_A | SCTLR_Axx_ELx_C |\ + SCTLR_Axx_ELx_WXN | SCTLR_Axx_ELx_EE) + +#if (SCTLR_EL2_SET ^SCTLR_EL2_CLEAR ) !=0xffffffffffffffffUL +#error "Inconsistent SCTLR_EL2 set/clear bits" +#endif + +#endif
Quoted 2 lines
xen/include/asm-arm/processor.h line 193
Quoted 2 lines
xen/include/asm-arm/processor.h +0
-0/* HCR Hyp Configuration Register */ #define HCR_RW (_AC(1,UL)<<31)/* Register Width, ARM64 only */
Cheers,
Signature · 6 lines
Julien Grall _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 04/19] xen/arm: Rework HSCTLR_BASE
Ping, it would be good to know which bits I dropped...
Quoted 184 lines · Julien Grall
Hi,
Quoted 10 lines · Stefano Stabellini
Quoted 9 lines · Julien Grall
The current value of
HSCTLR_BASEfor Arm64 is pretty wrong. It would actually turn on SCTLR_EL2.nAA (bit 6) on hardware implementing ARMv8.4-LSE.Furthermore, the documentation of what is cleared/set in
SCTLR_EL2is also not correct and looks like to be a verbatim copy from Arm32.
HSCTLR_BASEis replaced with a bunch of per-architecture new defines helping to understand better what is the initialie value fors/initialie/initial/
Quoted 5 lines
Quoted 4 lines
SCTLR_EL2/HSCTLR.
Note the defines
*_CLEARare only used to check the state of each bits are known.Quoted 8 lines
So basically the only purpose of
HSCTLR_CLEARis to execute:#if (
HSCTLR_SET^HSCTLR_CLEAR) !=0xffffffffURight? It is good to have the check.
Please add a one-line comment on top of
HSCTLR_CLEAR-- "only used to check that the state of each bits are known".We don't commonly add a comment every time a define is used only one time. So what's the benefits here?
In all honesty, such wording in the commit message was probably over the top. I am thinking to replace the sentence with:
"Lastly, all the bits are now described as either set or cleared. This allows us to check at pre-processing time the consistency of the initial value."
Quoted 85 lines
Quoted 85 lines
Lastly, the documentation is dropped from
arm{32,64}/head.Sas it would be pretty easy to get out-of-sync with the definitions.Signed-off-by Julien Grall <[email protected]>---
Changes in v2:
- UseBIT(..., UL)instead of_BITUL
---
xen/arch/arm/arm32/head.S | 12 +-------- xen/arch/arm/arm64/head.S | 10 +-------xen/include/asm-arm/processor.h| 54 ++++++++++++++++++++++++++++++++++++++++-3 files changed, 55 insertions(+), 21 deletions(-)xen/arch/arm/arm32/head.S +1
-11@@ -234,17 +234,7 @@ cpu_init_done: ldr r0, =(TCR_RES1|TCR_SH0_IS|TCR_ORGN0_WBWA|TCR_IRGN0_WBWA|TCR_T0SZ(0)) mcr CP32(r0, HTCR) - /* - * Set up the HSCTLR: - * Exceptions in LE ARM, - * Low-latency IRQs disabled, - * Write-implies-XN disabled (for now), - * D-cache disabled (for now), - * I-cache enabled, - * Alignment checking enabled, - * MMU translation disabled (for now). - */ - ldr r0, =(HSCTLR_BASE|SCTLR_Axx_ELx_A) + ldr r0, =HSCTLR_SET mcr CP32(r0, HSCTLR) /*xen/arch/arm/arm64/head.S +0
-0@@ -363,15 +363,7 @@ skip_bss:msr
tcr_el2, x0
- /* Set up theSCTLR_EL2:
- * Exceptions in LE ARM,
- * Low-latency IRQs disabled,
- * Write-implies-XN disabled (for now), - * D-cache disabled (for now),
- * I-cache enabled,
- * Alignment checking disabled,
- * MMU translation disabled (for now). */ - ldr x0, =(HSCTLR_BASE)
+ ldr x0, =SCTLR_EL2_SET
msrSCTLR_EL2, x0
/* Ensure that any exceptions encountered at EL2xen/include/asm-arm/processor.h +13
-1@@ -127,6 +127,9 @@ #define SCTLR_A32_ELx_TE BIT(30, UL) #define SCTLR_A32_ELx_FI BIT(21, UL) +/* Common bits for SCTLR_ELx for Arm64 */ +#define SCTLR_A64_ELx_SA BIT(3, UL) + /* Common bits for SCTLR_ELx on all architectures */ #define SCTLR_Axx_ELx_EE BIT(25, UL) #define SCTLR_Axx_ELx_WXN BIT(19, UL)@@ -135,7 +138,56 @@ #define SCTLR_Axx_ELx_A BIT(1, UL) #define SCTLR_Axx_ELx_M BIT(0, UL) -#define HSCTLR_BASE _AC(0x30c51878,U) +#ifdef CONFIG_ARM_32 + +#define HSCTLR_RES1 (BIT( 3, UL) | BIT( 4, UL) | BIT( 5, UL) |\ + BIT( 6, UL) | BIT(11, UL) | BIT(16, UL) |\ + BIT(18, UL) | BIT(22, UL) | BIT(23, UL) |\ + BIT(28, UL) | BIT(29, UL)) + +#define HSCTLR_RES0 (BIT(7, UL) | BIT(8, UL) | BIT(9, UL) | BIT(10, UL) |\ + BIT(13, UL) | BIT(14, UL) | BIT(15, UL) | BIT(17,UL ) |\ + BIT(20, UL) | BIT(24, UL) | BIT(26, UL) | BIT(27, UL) |\Quoted 8 lines
xen/include/asm-arm/processor.h line 154
Quoted 4 lines
xen/include/asm-arm/processor.h +4
-0+ BIT(31, UL)) + +/* Initial value for HSCTLR */ +#define HSCTLR_SET (HSCTLR_RES1 | SCTLR_Axx_ELx_A | SCTLR_Axx_ELx_I)As far as my calculations go, it looks like the only difference is
SCTLR_Axx_ELx_Acompared to0x30c51878, right? Which is the alignment check.That's correct and match the initial value on Arm32 (see
HSCTR_SET|SCTLR_Axx_ELx_Ain thehead.S).Quoted 23 lines
Quoted 23 lines
xen/include/asm-arm/processor.h +21
-0+#define HSCTLR_CLEAR (HSCTLR_RES0 | SCTLR_Axx_ELx_M |\ + SCTLR_Axx_ELx_C | SCTLR_Axx_ELx_WXN |\ + SCTLR_A32_ELx_FI | SCTLR_Axx_ELx_EE |\ + SCTLR_A32_ELx_TE) + +#if (HSCTLR_SET ^HSCTLR_CLEAR ) !=0xffffffffU +#error "Inconsistent HSCTLR set/clear bits" +#endif + +#else + +#define SCTLR_EL2_RES1 (BIT( 4, UL) | BIT( 5, UL) | BIT(11, UL) |\ + BIT(16, UL) | BIT(18, UL) | BIT(22, UL) |\ + BIT(23, UL) | BIT(28, UL) | BIT(29, UL)) + +#define SCTLR_EL2_RES0 (BIT( 6, UL) | BIT( 7, UL) | BIT( 8, UL) |\ + BIT( 9, UL) | BIT(10, UL) | BIT(13, UL) |\ + BIT(14, UL) | BIT(15, UL) | BIT(17, UL) |\ + BIT(20, UL) | BIT(21, UL) | BIT(24, UL) |\ + BIT(26, UL) | BIT(27, UL) | BIT(30, UL) |\ + BIT(31, UL) | (0xffffffffULL << 32))Quoted 8 lines
xen/include/asm-arm/processor.h line 180
Quoted 4 lines
xen/include/asm-arm/processor.h +4
-0+ +/* Initial value for SCTLR_EL2 */ +#define SCTLR_EL2_SET (SCTLR_EL2_RES1 | SCTLR_A64_ELx_SA |\ + SCTLR_Axx_ELx_I)Same here, you removed the reserved bits, and added the alignment check, same as for aarch32. If I got it right, it would be nice to add a statement like this to the commit message.
I don't see why "reserved bits" I dropped nor which alignment check I added.
It would be extremely useful if you provide more details in your review... In this case, it would be the exact bits I dropped/added.
Quoted 13 lines
Quoted 12 lines
xen/include/asm-arm/processor.h +9
-0+#define SCTLR_EL2_CLEAR (SCTLR_EL2_RES0 | SCTLR_Axx_ELx_M |\ + SCTLR_Axx_ELx_A | SCTLR_Axx_ELx_C |\ + SCTLR_Axx_ELx_WXN | SCTLR_Axx_ELx_EE) + +#if (SCTLR_EL2_SET ^SCTLR_EL2_CLEAR ) !=0xffffffffffffffffUL +#error "Inconsistent SCTLR_EL2 set/clear bits" +#endif + +#endif /* HCR Hyp Configuration Register */Quoted 1 line
#define HCR_RW (_AC(1,UL)<<31) /* Register Width, ARM64 only */
Cheers,
Signature · 6 lines
Julien Grall _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 04/19] xen/arm: Rework HSCTLR_BASE
Quoted 42 lines · Julien Grall
Ping, it would be good to know which bits I dropped...
Quoted 39 lines · Julien Grall
Hi,
Quoted 10 lines · Stefano Stabellini
Quoted 9 lines · Julien Grall
The current value of
HSCTLR_BASEfor Arm64 is pretty wrong. It would actually turn on SCTLR_EL2.nAA (bit 6) on hardware implementing ARMv8.4-LSE.Furthermore, the documentation of what is cleared/set in
SCTLR_EL2is also not correct and looks like to be a verbatim copy from Arm32.
HSCTLR_BASEis replaced with a bunch of per-architecture new defines helping to understand better what is the initialie value fors/initialie/initial/
Quoted 5 lines
Quoted 4 lines
SCTLR_EL2/HSCTLR.
Note the defines
*_CLEARare only used to check the state of each bits are known.Quoted 8 lines
So basically the only purpose of
HSCTLR_CLEARis to execute:#if (
HSCTLR_SET^HSCTLR_CLEAR) !=0xffffffffURight? It is good to have the check.
Please add a one-line comment on top of
HSCTLR_CLEAR-- "only used to check that the state of each bits are known".We don't commonly add a comment every time a define is used only one time. So what's the benefits here?
In all honesty, such wording in the commit message was probably over the top. I am thinking to replace the sentence with:
"Lastly, all the bits are now described as either set or cleared. This allows us to check at pre-processing time the consistency of the initial value."
This is even clearer, but I understood that part of the commit message well enough even before. I have no complaints there. My suggestion for an in-code comment is because the purpose of HSCTLR_CLEAR is not immediately obvious looking at the code only. The commit message is fine. I think that a one-liner in the code to say that HSCTLR_CLEAR is "only used at pre-processing time" would be good to have and beneficial for code readability.
Quoted 98 lines
Quoted 98 lines
Quoted 97 lines
Quoted 83 lines
Lastly, the documentation is dropped from
arm{32,64}/head.Sas it would be pretty easy to get out-of-sync with the definitions.Signed-off-by Julien Grall <[email protected]>---
Changes in v2:
- UseBIT(..., UL)instead of_BITUL
---
xen/arch/arm/arm32/head.S | 12 +--------
xen/arch/arm/arm64/head.S | 10 +-------
xen/include/asm-arm/processor.h| 54
++++++++++++++++++++++++++++++++++++++++-3 files changed, 55 insertions(+), 21 deletions(-)xen/arch/arm/arm32/head.S +1
-11@@ -234,17 +234,7 @@ cpu_init_done: ldr r0, =(TCR_RES1|TCR_SH0_IS|TCR_ORGN0_WBWA|TCR_IRGN0_WBWA|TCR_T0SZ(0)) mcr CP32(r0, HTCR) - /* - * Set up the HSCTLR: - * Exceptions in LE ARM, - * Low-latency IRQs disabled, - * Write-implies-XN disabled (for now), - * D-cache disabled (for now), - * I-cache enabled, - * Alignment checking enabled, - * MMU translation disabled (for now). - */ - ldr r0, =(HSCTLR_BASE|SCTLR_Axx_ELx_A) + ldr r0, =HSCTLR_SET mcr CP32(r0, HSCTLR) /*xen/arch/arm/arm64/head.S +0
-0@@ -363,15 +363,7 @@ skip_bss:msr
tcr_el2, x0
- /* Set up theSCTLR_EL2:
- * Exceptions in LE ARM,
- * Low-latency IRQs disabled,
- * Write-implies-XN disabled (for now), - * D-cache disabled (for now),
- * I-cache enabled,
- * Alignment checking disabled,
- * MMU translation disabled (for now). */ - ldr x0, =(HSCTLR_BASE)
+ ldr x0, =SCTLR_EL2_SET
msrSCTLR_EL2, x0
/* Ensure that any exceptions encountered at EL2xen/include/asm-arm/processor.h +11
-1@@ -127,6 +127,9 @@ #define SCTLR_A32_ELx_TE BIT(30, UL) #define SCTLR_A32_ELx_FI BIT(21, UL) +/* Common bits for SCTLR_ELx for Arm64 */ +#define SCTLR_A64_ELx_SA BIT(3, UL) + /* Common bits for SCTLR_ELx on all architectures */ #define SCTLR_Axx_ELx_EE BIT(25, UL) #define SCTLR_Axx_ELx_WXN BIT(19, UL)@@ -135,7 +138,56 @@ #define SCTLR_Axx_ELx_A BIT(1, UL) #define SCTLR_Axx_ELx_M BIT(0, UL) -#define HSCTLR_BASE _AC(0x30c51878,U) +#ifdef CONFIG_ARM_32 + +#define HSCTLR_RES1 (BIT( 3, UL) | BIT( 4, UL) | BIT( 5, UL) |\ + BIT( 6, UL) | BIT(11, UL) | BIT(16, UL) |\ + BIT(18, UL) | BIT(22, UL) | BIT(23, UL) |\ + BIT(28, UL) | BIT(29, UL)) + +#define HSCTLR_RES0 (BIT(7, UL) | BIT(8, UL) | BIT(9, UL) |xen/include/asm-arm/processor.h line 157
Quoted 10 lines
xen/include/asm-arm/processor.h +6
-0BIT(10, UL) |\ + BIT(13, UL) | BIT(14, UL) | BIT(15, UL) | BIT(17, UL) |\ + BIT(20, UL) | BIT(24, UL) | BIT(26, UL) | BIT(27, UL) |\ + BIT(31, UL)) + +/* Initial value for HSCTLR */ +#define HSCTLR_SET (HSCTLR_RES1 | SCTLR_Axx_ELx_A | SCTLR_Axx_ELx_I)As far as my calculations go, it looks like the only difference is
SCTLR_Axx_ELx_Acompared to0x30c51878, right? Which is the alignment check.
Quoted 2 lines
Quoted 2 lines
That's correct and match the initial value on Arm32 (see
HSCTR_SET|SCTLR_Axx_ELx_Ain thehead.S).
OK
Quoted 32 lines
Quoted 32 lines
Quoted 31 lines
Quoted 17 lines
xen/include/asm-arm/processor.h +15
-0+#define HSCTLR_CLEAR (HSCTLR_RES0 | SCTLR_Axx_ELx_M |\ + SCTLR_Axx_ELx_C | SCTLR_Axx_ELx_WXN |\ + SCTLR_A32_ELx_FI | SCTLR_Axx_ELx_EE |\ + SCTLR_A32_ELx_TE) + +#if (HSCTLR_SET ^HSCTLR_CLEAR ) !=0xffffffffU +#error "Inconsistent HSCTLR set/clear bits" +#endif + +#else + +#define SCTLR_EL2_RES1 (BIT( 4, UL) | BIT( 5, UL) | BIT(11, UL) |\ + BIT(16, UL) | BIT(18, UL) | BIT(22, UL) |\ + BIT(23, UL) | BIT(28, UL) | BIT(29, UL)) +xen/include/asm-arm/processor.h line 180
Quoted 10 lines
xen/include/asm-arm/processor.h +10
-0+#define SCTLR_EL2_RES0 (BIT( 6, UL) | BIT( 7, UL) | BIT( 8, UL) |\ + BIT( 9, UL) | BIT(10, UL) | BIT(13, UL) |\ + BIT(14, UL) | BIT(15, UL) | BIT(17, UL) |\ + BIT(20, UL) | BIT(21, UL) | BIT(24, UL) |\ + BIT(26, UL) | BIT(27, UL) | BIT(30, UL) |\ + BIT(31, UL) | (0xffffffffULL << 32)) + +/* Initial value for SCTLR_EL2 */ +#define SCTLR_EL2_SET (SCTLR_EL2_RES1 | SCTLR_A64_ELx_SA |\ + SCTLR_Axx_ELx_I)Same here, you removed the reserved bits, and added the alignment check, same as for aarch32. If I got it right, it would be nice to add a statement like this to the commit message.
Quoted 4 lines
Quoted 4 lines
I don't see why "reserved bits" I dropped nor which alignment check I added.
It would be extremely useful if you provide more details in your review... In this case, it would be the exact bits I dropped/added.
I looked at the full value of SCTLR_EL2_SET, it's 0x30c51838. I copy/paste here the wcalc command for our own convenience:
wcalc -h '(1<<4)|(1<<5)|(1<<11)|(1<<16)|(1<<18)|(1<<22)|(1<<23)|(1<<28)|(1<<29)|(1<<3)|(1<<12)'
HSCTLR_BASE is 0x30c51878. The difference is bit 6 which is RES0. It looks like I was wrong about the alignment check.
Quoted 14 lines
Quoted 14 lines
Quoted 14 lines
Quoted 11 lines
xen/include/asm-arm/processor.h +9
-0+#define SCTLR_EL2_CLEAR (SCTLR_EL2_RES0 | SCTLR_Axx_ELx_M |\ + SCTLR_Axx_ELx_A | SCTLR_Axx_ELx_C |\ + SCTLR_Axx_ELx_WXN | SCTLR_Axx_ELx_EE) + +#if (SCTLR_EL2_SET ^SCTLR_EL2_CLEAR ) !=0xffffffffffffffffUL +#error "Inconsistent SCTLR_EL2 set/clear bits" +#endif + +#endif Quoted 3 lines
/* HCR Hyp Configuration Register */
#define HCR_RW (_AC(1,UL)<<31) /* Register Width, ARM64 only */
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 04/19] xen/arm: Rework HSCTLR_BASE
Hi Stefano,
Quoted 44 lines · Stefano Stabellini
Quoted 42 lines · Julien Grall
Ping, it would be good to know which bits I dropped...
Quoted 39 lines · Julien Grall
Hi,
Quoted 10 lines · Stefano Stabellini
Quoted 9 lines · Julien Grall
The current value of
HSCTLR_BASEfor Arm64 is pretty wrong. It would actually turn on SCTLR_EL2.nAA (bit 6) on hardware implementing ARMv8.4-LSE.Furthermore, the documentation of what is cleared/set in
SCTLR_EL2is also not correct and looks like to be a verbatim copy from Arm32.
HSCTLR_BASEis replaced with a bunch of per-architecture new defines helping to understand better what is the initialie value fors/initialie/initial/
Quoted 5 lines
Quoted 4 lines
SCTLR_EL2/HSCTLR.
Note the defines
*_CLEARare only used to check the state of each bits are known.Quoted 8 lines
So basically the only purpose of
HSCTLR_CLEARis to execute:#if (
HSCTLR_SET^HSCTLR_CLEAR) !=0xffffffffURight? It is good to have the check.
Please add a one-line comment on top of
HSCTLR_CLEAR-- "only used to check that the state of each bits are known".We don't commonly add a comment every time a define is used only one time. So what's the benefits here?
In all honesty, such wording in the commit message was probably over the top. I am thinking to replace the sentence with:
"Lastly, all the bits are now described as either set or cleared. This allows us to check at pre-processing time the consistency of the initial value."
Quoted 7 lines
This is even clearer, but I understood that part of the commit message well enough even before. I have no complaints there. My suggestion for an in-code comment is because the purpose of
HSCTLR_CLEARis not immediately obvious looking at the code only. The commit message is fine. I think that a one-liner in the code to say thatHSCTLR_CLEARis "only used at pre-processing time" would be good to have and beneficial for code readability.
It is quite an odd comment as a lot of defines are only used for pre-processing it (i.e CONFIG_ or even macro generating function)... It is going to rot quickly but I can add it if you think it improves the code...
Quoted 9 lines
Quoted 8 lines
Quoted 8 lines
Quoted 3 lines
Same here, you removed the reserved bits, and added the alignment check, same as for aarch32. If I got it right, it would be nice to add a statement like this to the commit message.
I don't see why "reserved bits" I dropped nor which alignment check I added.
It would be extremely useful if you provide more details in your review... In this case, it would be the exact bits I dropped/added.
Quoted 7 lines
I looked at the full value of
SCTLR_EL2_SET, it's0x30c51838. I copy/paste here the wcalc command for our own convenience:wcalc -h '(1<<4)|(1<<5)|(1<<11)|(1<<16)|(1<<18)|(1<<22)|(1<<23)|(1<<28)|(1<<29)|(1<<3)|(1<<12)'
HSCTLR_BASEis0x30c51878. The difference is bit 6 which is RES0. It looks like I was wrong about the alignment check.
This was mentioned in the commit message:
"The current value of HSCTLR_BASE for Arm64 is pretty wrong. It would actually turn on SCTLR_EL2.nAA (bit 6) on hardware implementing ARMv8.4-LSE."
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 12 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
xen/include/asm-arm/processor.h +4
-0+#define SCTLR_EL2_CLEAR (SCTLR_EL2_RES0 | SCTLR_Axx_ELx_M |\ + SCTLR_Axx_ELx_A | SCTLR_Axx_ELx_C |\ + SCTLR_Axx_ELx_WXN | SCTLR_Axx_ELx_EE) +Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
xen/include/asm-arm/processor.h +5
-0+#if (SCTLR_EL2_SET ^SCTLR_EL2_CLEAR ) !=0xffffffffffffffffUL +#error "Inconsistent SCTLR_EL2 set/clear bits" +#endif + +#endif /* HCR Hyp Configuration Register */ #define HCR_RW (_AC(1,UL)<<31) /* Register Width, ARM64 only*/
Cheers,
Signature · 6 lines
Julien Grall _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 04/19] xen/arm: Rework HSCTLR_BASE
Quoted 58 lines · Julien Grall
Hi Stefano,
Quoted 47 lines · Stefano Stabellini
Quoted 45 lines · Julien Grall
Ping, it would be good to know which bits I dropped...
Quoted 42 lines · Julien Grall
Hi,
Quoted 11 lines · Stefano Stabellini
Quoted 10 lines · Julien Grall
The current value of
HSCTLR_BASEfor Arm64 is pretty wrong. It would actually turn on SCTLR_EL2.nAA (bit 6) on hardware implementing ARMv8.4-LSE.Furthermore, the documentation of what is cleared/set in
SCTLR_EL2is
also not correct and looks like to be a verbatim copy from Arm32.
HSCTLR_BASEis replaced with a bunch of per-architecture new defines helping to understand better what is the initialie value fors/initialie/initial/
Quoted 6 lines
Quoted 5 lines
SCTLR_EL2/HSCTLR.
Note the defines
*_CLEARare only used to check the state of each bits
are known.Quoted 8 lines
So basically the only purpose of
HSCTLR_CLEARis to execute:#if (
HSCTLR_SET^HSCTLR_CLEAR) !=0xffffffffURight? It is good to have the check.
Please add a one-line comment on top of
HSCTLR_CLEAR-- "only used to check that the state of each bits are known".We don't commonly add a comment every time a define is used only one time.
So what's the benefits here?In all honesty, such wording in the commit message was probably over the top. I am thinking to replace the sentence with:
"Lastly, all the bits are now described as either set or cleared. This allows us to check at pre-processing time the consistency of the initial value."
Quoted 7 lines
This is even clearer, but I understood that part of the commit message well enough even before. I have no complaints there. My suggestion for an in-code comment is because the purpose of
HSCTLR_CLEARis not immediately obvious looking at the code only. The commit message is fine. I think that a one-liner in the code to say thatHSCTLR_CLEARis "only used at pre-processing time" would be good to have and beneficial for code readability.
Quoted 3 lines
It is quite an odd comment as a lot of defines are only used for pre-processing it (i.e
CONFIG_or even macro generating function)... It is going to rot quickly but I can add it if you think it improves the code...
Yes, but this macro in particular is in the middle of other similarly named macros that are actually used at runtime. This is why I would like the comment. However, this is code readibility, and as you know it is a bit subjective.
Quoted 18 lines
Quoted 18 lines
Quoted 11 lines
Quoted 11 lines
Quoted 4 lines
Same here, you removed the reserved bits, and added the alignment check,
same as for aarch32. If I got it right, it would be nice to add a statement like this to the commit message.I don't see why "reserved bits" I dropped nor which alignment check I added.
It would be extremely useful if you provide more details in your review...
In this case, it would be the exact bits I dropped/added.I looked at the full value of
SCTLR_EL2_SET, it's0x30c51838. I copy/paste here the wcalc command for our own convenience:wcalc -h '(1<<4)|(1<<5)|(1<<11)|(1<<16)|(1<<18)|(1<<22)|(1<<23)|(1<<28)|(1<<29)|(1<<3)|(1<<12)'
Quoted 8 lines
Quoted 2 lines
HSCTLR_BASEis0x30c51878. The difference is bit 6 which is RES0. It looks like I was wrong about the alignment check.This was mentioned in the commit message:
"The current value of
HSCTLR_BASEfor Arm64 is pretty wrong. It would actually turn on SCTLR_EL2.nAA (bit 6) on hardware implementing ARMv8.4-LSE."
Good, it all checks out then.
Quoted 21 lines
Quoted 15 lines
Quoted 15 lines
Quoted 15 lines
Quoted 15 lines
Quoted 13 lines
xen/include/asm-arm/processor.h +9
-0+#define SCTLR_EL2_CLEAR (SCTLR_EL2_RES0 | SCTLR_Axx_ELx_M |\ + SCTLR_Axx_ELx_A | SCTLR_Axx_ELx_C |\ + SCTLR_Axx_ELx_WXN | SCTLR_Axx_ELx_EE) + +#if (SCTLR_EL2_SET ^SCTLR_EL2_CLEAR ) !=0xffffffffffffffffUL +#error "Inconsistent SCTLR_EL2 set/clear bits" +#endif + +#endif /* HCR Hyp Configuration Register */ #define HCR_RW (_AC(1,UL)<<31) /* Register Width, ARM64Quoted 2 lines
only
*/Cheers,
Signature · 1 line
Julien Grall
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Andrii Anisov
Reviewed-byRe: [PATCH MM-PART2 RESEND v2 04/19] xen/arm: Rework HSCTLR_BASE
In reply to Julien GrallQuoted 8 lines · Julien Grall
Hi,
Quoted 5 lines · Stefano Stabellini
Quoted 4 lines · Julien Grall
The current value of
HSCTLR_BASEfor Arm64 is pretty wrong. It would actually turn on SCTLR_EL2.nAA (bit 6) on hardware implementing ARMv8.4-LSE.
Quoted 12 lines
Quoted 5 lines
Quoted 5 lines
Furthermore, the documentation of what is cleared/set in
SCTLR_EL2is also not correct and looks like to be a verbatim copy from Arm32.
HSCTLR_BASEis replaced with a bunch of per-architecture new defines helping to understand better what is the initialie value fors/initialie/initial/
Quoted 4 lines
Quoted 4 lines
Lastly, the documentation is dropped from
arm{32,64}/head.Sas it would be pretty easy to get out-of-sync with the definitions.Signed-off-by Julien Grall <[email protected]>
FWIW, with misprint fixed
Signature · 7 lines
Sincerely, Andrii Anisov. _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 5/19 xen/arm: Remove parameter cpuid from start_xen
In reply to Julien GrallThe parameter cpuid is not used by start_xen. So remove it.
---
- Re-order the patch with "xen/arm: Rework secondary_start prototype"
--- xen/arch/arm/arm32/head.S | 1 - xen/arch/arm/arm64/head.S | 1 - xen/arch/arm/setup.c | 3 +-- 3 files changed, 1 insertion(+), 4 deletions(-)
xen/arch/arm/arm32/head.S +0 -1
@@ -447,7 +447,6 @@ launch: sub sp, #CPUINFO_sizeof /* Make room for CPU save record */ mov r0, r10 /* Marshal args: - phys_offset */ mov r1, r8 /* - DTB address */ - mov r2, r7 /* - CPU ID */ teq r12, #0 beq start_xen /* and disappear into the land of C */ b start_secondary /* (to the appropriate entry point) */
xen/arch/arm/arm64/head.S +0 -1
@@ -584,7 +584,6 @@ launch: mov x0, x20 /* Marshal args: - phys_offset */ mov x1, x21 /* - FDT */ - mov x2, x24 /* - CPU ID */ cbnz x22, 1f b start_xen /* and disappear into the land of C */ 1:
xen/arch/arm/setup.c +1 -2
@@ -733,8 +733,7 @@ size_t __read_mostly dcache_line_bytes;/* C entry point for boot CPU */ void __init start_xen (unsigned long boot_phys_offset, -unsigned long fdt_paddr, -unsigned long cpuid) +unsigned long fdt_paddr) {size_t fdt_size;int cpus, i;
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 05/19] xen/arm: Remove parameter cpuid from start_xen
Quoted 8 lines · Julien Grall
The parameter cpuid is not used by
start_xen. So remove it.Signed-off-by Julien Grall <[email protected]>---
- Re-order the patch with "xen/arm: Reworksecondary_startprototype"
---
Signature · 7 lines
Sincerely, Andrii Anisov. _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 05/19] xen/arm: Remove parameter cpuid from start_xen
Hi Stefano,
Gentle ping, you have acked the other patches but not this patch. I can't merge them without this one.
Cheers,
Quoted 14 lines · Andrii Anisov
Quoted 8 lines · Julien Grall
The parameter cpuid is not used by
start_xen. So remove it.Signed-off-by Julien Grall <[email protected]>---
- Re-order the patch with "xen/arm: Reworksecondary_startprototype"
---Reviewed-by Andrii Anisov <[email protected]>
Signature · 6 lines
Julien Grall _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 05/19] xen/arm: Remove parameter cpuid from start_xen
In reply to Julien GrallQuoted 3 lines · Julien Grall
The parameter cpuid is not used by
start_xen. So remove it.Signed-off-by Julien Grall <[email protected]>
Quoted 50 lines
---
- Re-order the patch with "xen/arm: Reworksecondary_startprototype"--- xen/arch/arm/arm32/head.S | 1 - xen/arch/arm/arm64/head.S | 1 - xen/arch/arm/setup.c | 3 +-- 3 files changed, 1 insertion(+), 4 deletions(-)xen/arch/arm/arm32/head.S +0
-1@@ -447,7 +447,6 @@ launch: sub sp, #CPUINFO_sizeof /* Make room for CPU save record */ mov r0, r10 /* Marshal args: - phys_offset */ mov r1, r8 /* - DTB address */ - mov r2, r7 /* - CPU ID */ teq r12, #0 beq start_xen /* and disappear into the land of C */ b start_secondary /* (to the appropriate entry point) */xen/arch/arm/arm64/head.S +0
-1@@ -584,7 +584,6 @@ launch: mov x0, x20 /* Marshal args: - phys_offset */ mov x1, x21 /* - FDT */ - mov x2, x24 /* - CPU ID */ cbnz x22, 1f b start_xen /* and disappear into the land of C */ 1:xen/arch/arm/setup.c +1
-2@@ -733,8 +733,7 @@ size_t __read_mostly dcache_line_bytes;/* C entry point for boot CPU */ void __init start_xen (unsigned long boot_phys_offset, -unsigned long fdt_paddr, -unsigned long cpuid) +unsigned long fdt_paddr) {size_t fdt_size;int cpus, i;Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 6/19 xen/arm: Rework secondary_start prototype
In reply to Julien GrallNone of the parameters of secondary_start are actually used. So turn secondary_start to a function with no parameters.
Also modify the assembly code to avoid setting-up the registers before calling secondary_start.
- Re-order the patch with "xen/arm: Remove parameter cpuid from start_xen".
--- xen/arch/arm/arm32/head.S | 4 ++-- xen/arch/arm/arm64/head.S | 3 ++- xen/arch/arm/smpboot.c | 4 +--- 3 files changed, 5 insertions(+), 6 deletions(-)
xen/arch/arm/arm32/head.S +2 -2
@@ -445,9 +445,9 @@ launch: ldr sp, [r0] add sp, #STACK_SIZE /* (which grows down from the top). */ sub sp, #CPUINFO_sizeof /* Make room for CPU save record */ - mov r0, r10 /* Marshal args: - phys_offset */ - mov r1, r8 /* - DTB address */ teq r12, #0 + moveq r0, r10 /* Marshal args: - phys_offset */ + moveq r1, r8 /* - DTB address */ beq start_xen /* and disappear into the land of C */ b start_secondary /* (to the appropriate entry point) */
xen/arch/arm/arm64/head.S +2 -1
@@ -582,9 +582,10 @@ launch: sub x0, x0, #CPUINFO_sizeof /* Make room for CPU save record */ mov sp, x0 + cbnz x22, 1f + mov x0, x20 /* Marshal args: - phys_offset */ mov x1, x21 /* - FDT */ - cbnz x22, 1f b start_xen /* and disappear into the land of C */ 1: b start_secondary /* (to the appropriate entry point) */
xen/arch/arm/smpboot.c +1 -3
@@ -297,9 +297,7 @@ smp_prepare_cpus(void) }/* Boot the current CPU */ -void start_secondary (unsigned long boot_phys_offset, -unsigned long fdt_paddr, -unsigned long hwid) +void start_secondary (void ) {unsigned int cpuid = init_data.cpuid;
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 06/19] xen/arm: Rework secondary_start prototype
Quoted 5 lines · Julien Grall
None of the parameters of
secondary_startare actually used. So turnsecondary_startto a function with no parameters.Also modify the assembly code to avoid setting-up the registers before calling
secondary_start.
It is called "start_secondary" rather than "secondary_start". Please fix the commit message. Then you can add
Quoted 60 lines
Signed-off-by Julien Grall <[email protected]>- Re-order the patch with "xen/arm: Remove parameter cpuid from
start_xen".--- xen/arch/arm/arm32/head.S | 4 ++-- xen/arch/arm/arm64/head.S | 3 ++- xen/arch/arm/smpboot.c | 4 +--- 3 files changed, 5 insertions(+), 6 deletions(-)xen/arch/arm/arm32/head.S +2
-2@@ -445,9 +445,9 @@ launch: ldr sp, [r0] add sp, #STACK_SIZE /* (which grows down from the top). */ sub sp, #CPUINFO_sizeof /* Make room for CPU save record */ - mov r0, r10 /* Marshal args: - phys_offset */ - mov r1, r8 /* - DTB address */ teq r12, #0 + moveq r0, r10 /* Marshal args: - phys_offset */ + moveq r1, r8 /* - DTB address */ beq start_xen /* and disappear into the land of C */ b start_secondary /* (to the appropriate entry point) */xen/arch/arm/arm64/head.S +2
-1@@ -582,9 +582,10 @@ launch: sub x0, x0, #CPUINFO_sizeof /* Make room for CPU save record */ mov sp, x0 + cbnz x22, 1f + mov x0, x20 /* Marshal args: - phys_offset */ mov x1, x21 /* - FDT */ - cbnz x22, 1f b start_xen /* and disappear into the land of C */ 1: b start_secondary /* (to the appropriate entry point) */xen/arch/arm/smpboot.c +1
-3@@ -297,9 +297,7 @@ smp_prepare_cpus(void) }/* Boot the current CPU */ -void start_secondary (unsigned long boot_phys_offset, -unsigned long fdt_paddr, -unsigned long hwid) +void start_secondary (void ) {unsigned int cpuid = init_data.cpuid;Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 06/19] xen/arm: Rework secondary_start prototype
Hi,
Quoted 9 lines · Stefano Stabellini
Quoted 5 lines · Julien Grall
None of the parameters of
secondary_startare actually used. So turnsecondary_startto a function with no parameters.Also modify the assembly code to avoid setting-up the registers before calling
secondary_start.It is called "
start_secondary" rather than "secondary_start". Please fix the commit message. Then you can add
Whoops, I will update it.
Quoted 5 lines
Reviewed-by Stefano Stabellini <[email protected]>Quoted 1 line
Signed-off-by Julien Grall <[email protected]>
I have also realized I forgot to add --- here.
Thank you,
Quoted 59 lines
Quoted 59 lines
- Re-order the patch with "xen/arm: Remove parameter cpuid from
start_xen".--- xen/arch/arm/arm32/head.S | 4 ++-- xen/arch/arm/arm64/head.S | 3 ++- xen/arch/arm/smpboot.c | 4 +--- 3 files changed, 5 insertions(+), 6 deletions(-)xen/arch/arm/arm32/head.S +2
-2@@ -445,9 +445,9 @@ launch: ldr sp, [r0] add sp, #STACK_SIZE /* (which grows down from the top). */ sub sp, #CPUINFO_sizeof /* Make room for CPU save record */ - mov r0, r10 /* Marshal args: - phys_offset */ - mov r1, r8 /* - DTB address */ teq r12, #0 + moveq r0, r10 /* Marshal args: - phys_offset */ + moveq r1, r8 /* - DTB address */ beq start_xen /* and disappear into the land of C */ b start_secondary /* (to the appropriate entry point) */xen/arch/arm/arm64/head.S +2
-1@@ -582,9 +582,10 @@ launch: sub x0, x0, #CPUINFO_sizeof /* Make room for CPU save record */ mov sp, x0 + cbnz x22, 1f + mov x0, x20 /* Marshal args: - phys_offset */ mov x1, x21 /* - FDT */ - cbnz x22, 1f b start_xen /* and disappear into the land of C */ 1: b start_secondary /* (to the appropriate entry point) */xen/arch/arm/smpboot.c +1
-3@@ -297,9 +297,7 @@ smp_prepare_cpus(void) }/* Boot the current CPU */ -void start_secondary (unsigned long boot_phys_offset, -unsigned long fdt_paddr, -unsigned long hwid) +void start_secondary (void ) {unsigned int cpuid = init_data.cpuid;Signature · 1 line
2.11.0
Signature · 6 lines
Julien Grall _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Andrii Anisov
Reviewed-byRe: [PATCH MM-PART2 RESEND v2 06/19] xen/arm: Rework secondary_start prototype
Quoted 7 lines · Julien Grall
Hi,
Quoted 4 lines · Stefano Stabellini
Quoted 3 lines · Julien Grall
None of the parameters of
secondary_startare actually used. So turnsecondary_startto a function with no parameters.
Quoted 13 lines
Quoted 5 lines
Quoted 2 lines
Also modify the assembly code to avoid setting-up the registers before calling
secondary_start.It is called "
start_secondary" rather than "secondary_start". Please fix the commit message. Then you can addWhoops, I will update it.
Quoted 5 lines
Reviewed-by Stefano Stabellini <[email protected]>Quoted 1 line
Signed-off-by Julien Grall <[email protected]>
FWIW, with the name fixed
Signature · 7 lines
Sincerely, Andrii Anisov. _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 7/19 xen/arm64: head: Remove unnecessary comment
In reply to Julien GrallSo far, we don't have specific core initialization at boot. So remove the comment.
---
Changes in v2:
- Fix typo in the commit message
- Add Andrii's reviewed-by
--- xen/arch/arm/arm64/head.S | 2 -- 1 file changed, 2 deletions(-)
xen/arch/arm/arm64/head.S +0 -2
@@ -344,8 +344,6 @@ el2: PRINT("- Xen starting at EL2 -\r\n") skip_bss: PRINT("- Setting up control registers -\r\n") - /* XXXX call PROCINFO_cpu_init here */ - /* Set up memory attribute type tables */ ldr x0, =MAIRVAL msr mair_el2, x0
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 07/19] xen/arm64: head: Remove unnecessary comment
Quoted 5 lines · Julien Grall
So far, we don't have specific core initialization at boot. So remove the comment.
Signed-off-by Julien Grall <[email protected]>Reviewed-by Andrii Anisov <[email protected]>
Quoted 24 lines
---
Changes in v2:
- Fix typo in the commit message
- Add Andrii's reviewed-by--- xen/arch/arm/arm64/head.S | 2 -- 1 file changed, 2 deletions(-)xen/arch/arm/arm64/head.S +0
-2@@ -344,8 +344,6 @@ el2: PRINT("- Xen starting at EL2 -\r\n") skip_bss: PRINT("- Setting up control registers -\r\n") - /* XXXX call PROCINFO_cpu_init here */ - /* Set up memory attribute type tables */ ldr x0, =MAIRVAL msr mair_el2, x0Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 8/19 xen/arm64: head: Move earlyprintk messages in .rodata.str
In reply to Julien GrallAt the moment, the earlyprintk messages are interleaved with the instructions. This makes more difficult to read the objdump output.
Introduce a new macro to add a string in .rodata.str and use it for all the earlyprintk messages.
---
I haven't done a similar change in arm32 yet because the compiler will throw an error when using 'adr' when load an address from a different section (see A5-200 in ARM DDI 0406C.a for the technical reason). The change is likely to be more elaborate.
Changes in v2:
- Add Andrii's reviewed-by
--- xen/arch/arm/arm64/head.S | 14 +++++--------- xen/include/asm-arm/asm_defns.h | 5 +++++ 2 files changed, 10 insertions(+), 9 deletions(-)
xen/arch/arm/arm64/head.S +5 -9
@@ -81,13 +81,10 @@ /* Macro to print a string to the UART, if there is one. * Clobbers x0-x3. */ #ifdef CONFIG_EARLY_PRINTK -#define PRINT(_s) \ - adr x0, 98f ; \ - bl puts ; \ - b 99f ; \ -98: .asciz _s ; \ - .align 2 ; \ -99: +#define PRINT(_s) \ + adr x0, 98f ; \ + bl puts ; \ + RODATA_STR(98, _s) #else /* CONFIG_EARLY_PRINTK */ #define PRINT(s) #endif /* !CONFIG_EARLY_PRINTK */
@@ -633,8 +630,7 @@ init_uart: #endif adr x0, 1f b puts -1: .asciz "- UART enabled -\r\n" - .align 4 +RODATA_STR(1, "- UART enabled -\r\n") /* Print early debug messages. * x0: Nul-terminated string to print.
xen/include/asm-arm/asm_defns.h +5 -0
@@ -16,6 +16,11 @@# error "unknown ARM variant"#endif +#define RODATA_STR (label, msg) \ +.pushsection .rodata.str, "aMS", %progbits, 1 ; \ +label: .asciz msg; \ +.popsection +#endif /* __ARM_ASM_DEFNS_H__ */ /* * Local variables:
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 08/19] xen/arm64: head: Move earlyprintk messages in .rodata.str
Quoted 8 lines · Julien Grall
At the moment, the earlyprintk messages are interleaved with the instructions. This makes more difficult to read the objdump output.
Introduce a new macro to add a string in .rodata.str and use it for all the earlyprintk messages.
Signed-off-by Julien Grall <[email protected]>Reviewed-by Andrii Anisov <[email protected]>
Quoted 65 lines
---
I haven't done a similar change in arm32 yet because the compiler will throw an error when using 'adr' when load an address from a different section (see A5-200 in ARM DDI 0406C.a for the technical reason). The change is likely to be more elaborate.
Changes in v2:
- Add Andrii's reviewed-by--- xen/arch/arm/arm64/head.S | 14 +++++--------- xen/include/asm-arm/asm_defns.h | 5 +++++ 2 files changed, 10 insertions(+), 9 deletions(-)xen/arch/arm/arm64/head.S +5
-9@@ -81,13 +81,10 @@ /* Macro to print a string to the UART, if there is one. * Clobbers x0-x3. */ #ifdef CONFIG_EARLY_PRINTK -#define PRINT(_s) \ - adr x0, 98f ; \ - bl puts ; \ - b 99f ; \ -98: .asciz _s ; \ - .align 2 ; \ -99: +#define PRINT(_s) \ + adr x0, 98f ; \ + bl puts ; \ + RODATA_STR(98, _s) #else /* CONFIG_EARLY_PRINTK */ #define PRINT(s) #endif /* !CONFIG_EARLY_PRINTK */@@ -633,8 +630,7 @@ init_uart: #endif adr x0, 1f b puts -1: .asciz "- UART enabled -\r\n" - .align 4 +RODATA_STR(1, "- UART enabled -\r\n") /* Print early debug messages. * x0: Nul-terminated string to print.xen/include/asm-arm/asm_defns.h +5
-0@@ -16,6 +16,11 @@# error "unknown ARM variant"#endif +#define RODATA_STR (label, msg) \ +.pushsection .rodata.str, "aMS", %progbits, 1 ; \ +label: .asciz msg; \ +.popsection +#endif /* __ARM_ASM_DEFNS_H__ */ /* * Local variables: Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 9/19 xen/arm64: head: Correctly report the HW CPU ID
In reply to Julien GrallThere are no reason to consider the HW CPU ID will be 0 when the processor is part of a uniprocessor system. At best, this will result to conflicting output as the rest of Xen use the value directly read from MPIDR_EL1.
So remove the zeroing and logic to check if the CPU is part of a uniprocessor system.
---
Changes in v2:
- Add Andrii's reviewed-by
--- xen/arch/arm/arm64/head.S | 6 ------ 1 file changed, 6 deletions(-)
xen/arch/arm/arm64/head.S +0 -6
@@ -277,15 +277,9 @@ GLOBAL(init_secondary) mov x26, #1 /* X26 := skip_zero_bss */ common_start: - mov x24, #0 /* x24 := CPU ID. Initialy zero until we - * find that multiprocessor extensions are - * present and the system is SMP */ mrs x0, mpidr_el1 - tbnz x0, _MPIDR_UP, 1f /* Uniprocessor system? */ - ldr x13, =(~MPIDR_HWID_MASK) bic x24, x0, x13 /* Mask out flags to get CPU ID */ -1: /* Non-boot CPUs wait here until __cpu_up is ready for them */ cbz x22, 1f
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 09/19] xen/arm64: head: Correctly report the HW CPU ID
Quoted 10 lines · Julien Grall
There are no reason to consider the HW CPU ID will be 0 when the processor is part of a uniprocessor system. At best, this will result to conflicting output as the rest of Xen use the value directly read from
MPIDR_EL1.So remove the zeroing and logic to check if the CPU is part of a uniprocessor system.
Signed-off-by Julien Grall <[email protected]>Reviewed-by Andrii Anisov <[email protected]>
Quoted 30 lines
---
Changes in v2:
- Add Andrii's reviewed-by--- xen/arch/arm/arm64/head.S | 6 ------ 1 file changed, 6 deletions(-)xen/arch/arm/arm64/head.S +0
-6@@ -277,15 +277,9 @@ GLOBAL(init_secondary) mov x26, #1 /* X26 := skip_zero_bss */ common_start: - mov x24, #0 /* x24 := CPU ID. Initialy zero until we - * find that multiprocessor extensions are - * present and the system is SMP */ mrs x0, mpidr_el1 - tbnz x0, _MPIDR_UP, 1f /* Uniprocessor system? */ - ldr x13, =(~MPIDR_HWID_MASK) bic x24, x0, x13 /* Mask out flags to get CPU ID */ -1: /* Non-boot CPUs wait here until __cpu_up is ready for them */ cbz x22, 1fSignature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 10/19 xen/arm32: head: Correctly report the HW CPU ID
In reply to Julien GrallThere are no reason to consider the HW CPU ID will be 0 when the processor is part of a uniprocessor system. At best, this will result to conflicting output as the rest of Xen use the value directly read from MPIDR.
So remove the zeroing and logic to check if the CPU is part of a uniprocessor system.
---
Changes in v2:
- Add Andrii's reviewed-by
--- xen/arch/arm/arm32/head.S | 8 -------- 1 file changed, 8 deletions(-)
xen/arch/arm/arm32/head.S +0 -8
@@ -124,16 +124,8 @@ GLOBAL(init_secondary) mov r12, #1 /* r12 := is_secondary_cpu */ common_start: - mov r7, #0 /* r7 := CPU ID. Initialy zero until we - * find that multiprocessor extensions are - * present and the system is SMP */ mrc CP32(r1, MPIDR) - tst r1, #MPIDR_SMP /* Multiprocessor extension supported? */ - beq 1f - tst r1, #MPIDR_UP /* Uniprocessor system? */ - bne 1f bic r7, r1, #(~MPIDR_HWID_MASK) /* Mask out flags to get CPU ID */ -1: /* Non-boot CPUs wait here until __cpu_up is ready for them */ teq r12, #0
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 10/19] xen/arm32: head: Correctly report the HW CPU ID
Quoted 10 lines · Julien Grall
There are no reason to consider the HW CPU ID will be 0 when the processor is part of a uniprocessor system. At best, this will result to conflicting output as the rest of Xen use the value directly read from MPIDR.
So remove the zeroing and logic to check if the CPU is part of a uniprocessor system.
Signed-off-by Julien Grall <[email protected]>Reviewed-by Andrii Anisov <[email protected]>
Quoted 31 lines
---
Changes in v2:
- Add Andrii's reviewed-by--- xen/arch/arm/arm32/head.S | 8 -------- 1 file changed, 8 deletions(-)xen/arch/arm/arm32/head.S +0
-8@@ -124,16 +124,8 @@ GLOBAL(init_secondary) mov r12, #1 /* r12 := is_secondary_cpu */ common_start: - mov r7, #0 /* r7 := CPU ID. Initialy zero until we - * find that multiprocessor extensions are - * present and the system is SMP */ mrc CP32(r1, MPIDR) - tst r1, #MPIDR_SMP /* Multiprocessor extension supported? */ - beq 1f - tst r1, #MPIDR_UP /* Uniprocessor system? */ - bne 1f bic r7, r1, #(~MPIDR_HWID_MASK) /* Mask out flags to get CPU ID */ -1: /* Non-boot CPUs wait here until __cpu_up is ready for them */ teq r12, #0Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 11/19 xen/arm32: head: Don't set MAIR0 and MAIR1
In reply to Julien GrallThe co-processor registers MAIR0 and MAIR1 are managed by EL1. So there are no need to initialize them during Xen boot.
---
Changes in v2
- Add Andrii's reviewed-by
--- xen/arch/arm/arm32/head.S | 2 -- 1 file changed, 2 deletions(-)
xen/arch/arm/arm32/head.S +0 -2
@@ -212,8 +212,6 @@ cpu_init_done: /* Set up memory attribute type tables */ ldr r0, =MAIR0VAL ldr r1, =MAIR1VAL - mcr CP32(r0, MAIR0) - mcr CP32(r1, MAIR1) mcr CP32(r0, HMAIR0) mcr CP32(r1, HMAIR1)
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 11/19] xen/arm32: head: Don't set MAIR0 and MAIR1
Quoted 5 lines · Julien Grall
The co-processor registers MAIR0 and MAIR1 are managed by EL1. So there are no need to initialize them during Xen boot.
Signed-off-by Julien Grall <[email protected]>Reviewed-by Andrii Anisov <[email protected]>
Quoted 23 lines
---
Changes in v2
- Add Andrii's reviewed-by--- xen/arch/arm/arm32/head.S | 2 -- 1 file changed, 2 deletions(-)xen/arch/arm/arm32/head.S +0
-2@@ -212,8 +212,6 @@ cpu_init_done: /* Set up memory attribute type tables */ ldr r0, =MAIR0VAL ldr r1, =MAIR1VAL - mcr CP32(r0, MAIR0) - mcr CP32(r1, MAIR1) mcr CP32(r0, HMAIR0) mcr CP32(r1, HMAIR1)Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 12/19 xen/arm32: head: Always zero r3 before update a page-table entry
In reply to Julien GrallThe boot code is using r2 and r3 to hold the page-table entry value. While r2 is always updated before storing the value, this is not always the case for r3.
Thankfully today, r3 will always be zero when we care. But this is difficult to track and error-prone.
So always zero r3 within the few instructions before the write the page-table entry.
---
Changes in v2:
- Use 0x0 instead of 0
- Remove a duplicate mov r3, #0
--- xen/arch/arm/arm32/head.S | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
xen/arch/arm/arm32/head.S +3 -1
@@ -270,6 +270,7 @@ cpu_init_done: orr r2, r2, #PT_UPPER(MEM) /* r2:r3 := section map */ orr r2, r2, #PT_LOWER(MEM) lsl r1, r1, #3 /* r1 := Slot offset */ + mov r3, #0x0 strd r2, r3, [r4, r1] /* Mapping of paddr(start) */ mov r6, #1 /* r6 := identity map now in place */
@@ -372,11 +373,11 @@ paging: /* Add UART to the fixmap table */ ldr r1, =xen_fixmap /* r1 := vaddr (xen_fixmap) */ - mov r3, #0 lsr r2, r11, #THIRD_SHIFT lsl r2, r2, #THIRD_SHIFT /* 4K aligned paddr of UART */ orr r2, r2, #PT_UPPER(DEV_L3) orr r2, r2, #PT_LOWER(DEV_L3) /* r2:r3 := 4K dev map including UART */ + mov r3, #0x0 strd r2, r3, [r1, #(FIXMAP_CONSOLE*8)] /* Map it in the first fixmap's slot */ 1:
@@ -388,6 +389,7 @@ paging: orr r2, r2, #PT_LOWER(PT) /* r2:r3 := table map of xen_fixmap */ ldr r4, =FIXMAP_ADDR(0) mov r4, r4, lsr #(SECOND_SHIFT - 3) /* r4 := Slot for FIXMAP(0) */ + mov r3, #0x0 strd r2, r3, [r1, r4] /* Map it in the fixmap's slot */ /* Use a virtual address to access the UART. */
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 12/19] xen/arm32: head: Always zero r3 before update a page-table entry
Quoted 10 lines · Julien Grall
The boot code is using r2 and r3 to hold the page-table entry value. While r2 is always updated before storing the value, this is not always the case for r3.
Thankfully today, r3 will always be zero when we care. But this is difficult to track and error-prone.
So always zero r3 within the few instructions before the write the page-table entry.
Quoted 7 lines
Signed-off-by Julien Grall <[email protected]>---
Changes in v2:
- Use0x0instead of 0
- Remove a duplicate mov r3, #0
---
Signature · 7 lines
Sincerely, Andrii Anisov. _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 12/19] xen/arm32: head: Always zero r3 before update a page-table entry
In reply to Julien GrallQuoted 11 lines · Julien Grall
The boot code is using r2 and r3 to hold the page-table entry value. While r2 is always updated before storing the value, this is not always the case for r3.
Thankfully today, r3 will always be zero when we care. But this is difficult to track and error-prone.
So always zero r3 within the few instructions before the write the page-table entry.
Signed-off-by Julien Grall <[email protected]>
Quoted 44 lines
---
Changes in v2:
- Use0x0instead of 0
- Remove a duplicate mov r3, #0--- xen/arch/arm/arm32/head.S | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)xen/arch/arm/arm32/head.S +3
-1@@ -270,6 +270,7 @@ cpu_init_done: orr r2, r2, #PT_UPPER(MEM) /* r2:r3 := section map */ orr r2, r2, #PT_LOWER(MEM) lsl r1, r1, #3 /* r1 := Slot offset */ + mov r3, #0x0 strd r2, r3, [r4, r1] /* Mapping of paddr(start) */ mov r6, #1 /* r6 := identity map now in place */@@ -372,11 +373,11 @@ paging: /* Add UART to the fixmap table */ ldr r1, =xen_fixmap /* r1 := vaddr (xen_fixmap) */ - mov r3, #0 lsr r2, r11, #THIRD_SHIFT lsl r2, r2, #THIRD_SHIFT /* 4K aligned paddr of UART */ orr r2, r2, #PT_UPPER(DEV_L3) orr r2, r2, #PT_LOWER(DEV_L3) /* r2:r3 := 4K dev map including UART */ + mov r3, #0x0 strd r2, r3, [r1, #(FIXMAP_CONSOLE*8)] /* Map it in the first fixmap's slot */ 1:@@ -388,6 +389,7 @@ paging: orr r2, r2, #PT_LOWER(PT) /* r2:r3 := table map of xen_fixmap */ ldr r4, =FIXMAP_ADDR(0) mov r4, r4, lsr #(SECOND_SHIFT - 3) /* r4 := Slot for FIXMAP(0) */ + mov r3, #0x0 strd r2, r3, [r1, r4] /* Map it in the fixmap's slot */ /* Use a virtual address to access the UART. */Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 13/19 xen/arm32: mm: Avoid to zero and clean cache for CPU0 domheap
In reply to Julien GrallThe page-table walker is configured to use the same shareability and cacheability as the access performed when updating the page-tables. This means cleaning the cache for CPU0 domheap is unnecessary.
Furthermore, CPU0 page-tables are part of Xen binary and will already be zeroed before been used. So it is pointless to zero the domheap again.
---
Changes in v2:
- Tweak a bit the commit message
- Add Andrii's reviewed-by
--- xen/arch/arm/mm.c | 5 ----- 1 file changed, 5 deletions(-)
xen/arch/arm/mm.c +0 -5
@@ -724,11 +724,6 @@ void __init setup_pagetables(unsigned long boot_phys_offset)#ifdef CONFIG_ARM_32 per_cpu (xen_pgtable,0 ) = cpu0_pgtable;per_cpu (xen_dommap,0 ) = cpu0_dommap; - -/* Make sure it is clear */ -memset (this_cpu (xen_dommap),0 ,DOMHEAP_SECOND_PAGES *PAGE_SIZE ); -clean_dcache_va_range (this_cpu (xen_dommap), -DOMHEAP_SECOND_PAGES *PAGE_SIZE );#endif }
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 13/19] xen/arm32: mm: Avoid to zero and clean cache for CPU0 domheap
Quoted 9 lines · Julien Grall
The page-table walker is configured to use the same shareability and cacheability as the access performed when updating the page-tables. This means cleaning the cache for CPU0 domheap is unnecessary.
Furthermore, CPU0 page-tables are part of Xen binary and will already be zeroed before been used. So it is pointless to zero the domheap again.
Signed-off-by Julien Grall <[email protected]>Reviewed-by Andrii Anisov <[email protected]>
Quoted 27 lines
---
Changes in v2:
- Tweak a bit the commit message
- Add Andrii's reviewed-by--- xen/arch/arm/mm.c | 5 ----- 1 file changed, 5 deletions(-)xen/arch/arm/mm.c +0
-5@@ -724,11 +724,6 @@ void __init setup_pagetables(unsigned long boot_phys_offset)#ifdef CONFIG_ARM_32 per_cpu (xen_pgtable,0 ) = cpu0_pgtable;per_cpu (xen_dommap,0 ) = cpu0_dommap; - -/* Make sure it is clear */ -memset (this_cpu (xen_dommap),0 ,DOMHEAP_SECOND_PAGES *PAGE_SIZE ); -clean_dcache_va_range (this_cpu (xen_dommap), -DOMHEAP_SECOND_PAGES *PAGE_SIZE );#endif }Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 14/19 xen/arm32: mm: Avoid cleaning the cache for secondary CPUs page-tables
In reply to Julien GrallThe page-table walker is configured to use the same shareability and cacheability as the access performed when updating the page-tables. This means cleaning the cache for secondary CPUs runtime page-tables is unnecessary.
---
Changes in v2:
- Add Andrii's reviewed-by
--- xen/arch/arm/mm.c | 3 --- 1 file changed, 3 deletions(-)
xen/arch/arm/mm.c +0 -3
@@ -769,9 +769,6 @@ int init_secondary_pagetables(int cpu)write_pte (&first[first_table_offset (DOMHEAP_VIRT_START +i*FIRST_SIZE )], pte); } -clean_dcache_va_range (first,PAGE_SIZE ); -clean_dcache_va_range (domheap,DOMHEAP_SECOND_PAGES *PAGE_SIZE ); -per_cpu (xen_pgtable, cpu) = first;per_cpu (xen_dommap, cpu) = domheap;
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 14/19] xen/arm32: mm: Avoid cleaning the cache for secondary CPUs page-tables
Quoted 4 lines · Julien Grall
The page-table walker is configured to use the same shareability and cacheability as the access performed when updating the page-tables. This means cleaning the cache for secondary CPUs runtime page-tables is unnecessary.
All right. Is there an explicit configuration for the shareability and cacheability used by the page-table walker or is it specified as such in the Arm Arm? Also, isn't it possible that CPUs on a different cluster (big.LITTLE) would have issues with this if the cache could be split between the two clusters?
Quoted 27 lines
Signed-off-by Julien Grall <[email protected]>Reviewed-by Andrii Anisov <[email protected]>---
Changes in v2:
- Add Andrii's reviewed-by--- xen/arch/arm/mm.c | 3 --- 1 file changed, 3 deletions(-)xen/arch/arm/mm.c +0
-3@@ -769,9 +769,6 @@ int init_secondary_pagetables(int cpu)write_pte (&first[first_table_offset (DOMHEAP_VIRT_START +i*FIRST_SIZE )], pte); } -clean_dcache_va_range (first,PAGE_SIZE ); -clean_dcache_va_range (domheap,DOMHEAP_SECOND_PAGES *PAGE_SIZE ); -per_cpu (xen_pgtable, cpu) = first;per_cpu (xen_dommap, cpu) = domheap;Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 14/19] xen/arm32: mm: Avoid cleaning the cache for secondary CPUs page-tables
Quoted 9 lines · Stefano Stabellini
Quoted 4 lines · Julien Grall
The page-table walker is configured to use the same shareability and cacheability as the access performed when updating the page-tables. This means cleaning the cache for secondary CPUs runtime page-tables is unnecessary.
All right. Is there an explicit configuration for the shareability and cacheability used by the page-table walker or is it specified as such in the Arm Arm?
See the configuration of TCR_EL2, I can mention it.
Quoted 3 lines
Also, isn't it possible that CPUs on a different cluster (big.LITTLE) would have issues with this if the cache could be split between the two clusters?
I don't understand this... Cache should be coherent when a CPU leaves EL3.
But we already share some bits of the page tables between the processor (see create_xen_page_tables). So I don't see where there is a possible problem here.
Cheers,
Quoted 29 lines
Quoted 29 lines
Signed-off-by Julien Grall <[email protected]>Reviewed-by Andrii Anisov <[email protected]>---
Changes in v2:
- Add Andrii's reviewed-by--- xen/arch/arm/mm.c | 3 --- 1 file changed, 3 deletions(-)xen/arch/arm/mm.c +0
-3@@ -769,9 +769,6 @@ int init_secondary_pagetables(int cpu)write_pte (&first[first_table_offset (DOMHEAP_VIRT_START +i*FIRST_SIZE )], pte); } -clean_dcache_va_range (first,PAGE_SIZE ); -clean_dcache_va_range (domheap,DOMHEAP_SECOND_PAGES *PAGE_SIZE ); -per_cpu (xen_pgtable, cpu) = first;per_cpu (xen_dommap, cpu) = domheap;Signature · 1 line
2.11.0
Signature · 6 lines
Julien Grall _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 14/19] xen/arm32: mm: Avoid cleaning the cache for secondary CPUs page-tables
Hi,
Quoted 7 lines · Julien Grall
Quoted 6 lines · Stefano Stabellini
Quoted 4 lines · Julien Grall
The page-table walker is configured to use the same shareability and cacheability as the access performed when updating the page-tables. This means cleaning the cache for secondary CPUs runtime page-tables is unnecessary.
Quoted 9 lines
Quoted 3 lines
All right. Is there an explicit configuration for the shareability and cacheability used by the page-table walker or is it specified as such in the Arm Arm?
See the configuration of
TCR_EL2, I can mention it.Quoted 3 lines
Also, isn't it possible that CPUs on a different cluster (big.LITTLE) would have issues with this if the cache could be split between the two clusters?
Gentle ping, can you please clarify your question.
Quoted 3 lines
I don't understand this... Cache should be coherent when a CPU leaves EL3.
But we already share some bits of the page tables between the processor (seecreate_xen_page_tables). So I don't see where there is a possible problem here.
Cheers,
Signature · 6 lines
Julien Grall _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 14/19] xen/arm32: mm: Avoid cleaning the cache for secondary CPUs page-tables
In reply to Julien GrallQuoted 7 lines · Julien Grall
Quoted 6 lines · Stefano Stabellini
Quoted 4 lines · Julien Grall
The page-table walker is configured to use the same shareability and cacheability as the access performed when updating the page-tables. This means cleaning the cache for secondary CPUs runtime page-tables is unnecessary.
Quoted 5 lines
Quoted 3 lines
All right. Is there an explicit configuration for the shareability and cacheability used by the page-table walker or is it specified as such in the Arm Arm?
See the configuration of
TCR_EL2, I can mention it.
That would be nice. I double-checked and it is as you wrote.
Quoted 8 lines
Quoted 3 lines
Also, isn't it possible that CPUs on a different cluster (big.LITTLE) would have issues with this if the cache could be split between the two clusters?
I don't understand this... Cache should be coherent when a CPU leaves EL3. But we already share some bits of the page tables between the processor (see
create_xen_page_tables). So I don't see where there is a possible problem here.
If the cache is always coherent across the clusters and the pagetable-walkers of different clusters, then this is fine.
Quoted 24 lines
Quoted 24 lines
Quoted 24 lines
Signed-off-by Julien Grall <[email protected]>Reviewed-by Andrii Anisov <[email protected]>---
Changes in v2:
- Add Andrii's reviewed-by--- xen/arch/arm/mm.c | 3 --- 1 file changed, 3 deletions(-)xen/arch/arm/mm.c +0
-2@@ -769,9 +769,6 @@ int init_secondary_pagetables(int cpu)write_pte (&first[first_table_offset (DOMHEAP_VIRT_START +i*FIRST_SIZE )], pte); } -clean_dcache_va_range (first,PAGE_SIZE ); -clean_dcache_va_range (domheap,DOMHEAP_SECOND_PAGES *PAGE_SIZE ); -per_cpu (xen_pgtable, cpu) = first;per_cpu (xen_dommap, cpu) = domheap;
Quoted 7 lines
Quoted 3 lines
Quoted 3 lines
Signature · 1 line
2.11.0Signature · 1 line
Julien Grall
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 14/19] xen/arm32: mm: Avoid cleaning the cache for secondary CPUs page-tables
Hi Steano,
Quoted 8 lines · Stefano Stabellini
Quoted 7 lines · Julien Grall
Quoted 6 lines · Stefano Stabellini
Quoted 4 lines · Julien Grall
The page-table walker is configured to use the same shareability and cacheability as the access performed when updating the page-tables. This means cleaning the cache for secondary CPUs runtime page-tables is unnecessary.
Quoted 7 lines
Quoted 5 lines
Quoted 3 lines
All right. Is there an explicit configuration for the shareability and cacheability used by the page-table walker or is it specified as such in the Arm Arm?
See the configuration of
TCR_EL2, I can mention it.That would be nice. I double-checked and it is as you wrote.
Sure. How about:
"The page-table walker is configured by TCR_EL2 to use shareability and cacheability as the access performed when updating the page-tables. [...]"
Quoted 6 lines
Quoted 6 lines
Quoted 5 lines
Also, isn't it possible that CPUs on a different cluster (big.LITTLE) would have issues with this if the cache could be split between the two clusters?
Quoted 7 lines
Quoted 4 lines
I don't understand this... Cache should be coherent when a CPU leaves EL3. But we already share some bits of the page tables between the processor (see
create_xen_page_tables). So I don't see where there is a possible problem here.If the cache is always coherent across the clusters and the pagetable-walkers of different clusters, then this is fine.
Xen (and Linux) built on the assumption that all the CPUs (and page-table walker) are in the same shareable domain (i.e innershareable). If you have a platform where it is not the case, then Xen is going to be badly broken.
This is also inline with the expectatio from the Arm Arm (B2-123 in DDI 0487D.a):
"The Inner Shareable domain is expected to be the set of PEs controlled by a single hypervisor or operating system."
Do you have a case where Xen needs to run on PEs in different domains?
Cheers,
Signature · 6 lines
Julien Grall _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 14/19] xen/arm32: mm: Avoid cleaning the cache for secondary CPUs page-tables
Quoted 16 lines · Julien Grall
Hi Steano,
Quoted 13 lines · Stefano Stabellini
Quoted 12 lines · Julien Grall
Quoted 10 lines · Stefano Stabellini
Quoted 5 lines · Julien Grall
The page-table walker is configured to use the same shareability and cacheability as the access performed when updating the page-tables. This
means cleaning the cache for secondary CPUs runtime page-tables is unnecessary.All right. Is there an explicit configuration for the shareability and cacheability used by the page-table walker or is it specified as such in the Arm Arm?
Quoted 7 lines
Quoted 2 lines
Quoted 1 line
See the configuration of
TCR_EL2, I can mention it.That would be nice. I double-checked and it is as you wrote.
Sure. How about:
"The page-table walker is configured by
TCR_EL2to use shareability and cacheability as the access performed when updating the page-tables. [...]"
That's great thank you. With that, add my reviewed-by.
Quoted 17 lines
Quoted 4 lines
Quoted 4 lines
Quoted 3 lines
Also, isn't it possible that CPUs on a different cluster (big.LITTLE) would have issues with this if the cache could be split between the two clusters?
Quoted 8 lines
Quoted 5 lines
I don't understand this... Cache should be coherent when a CPU leaves EL3. But we already share some bits of the page tables between the processor (see
create_xen_page_tables). So I don't see where there is a possible problem here.If the cache is always coherent across the clusters and the pagetable-walkers of different clusters, then this is fine.
Xen (and Linux) built on the assumption that all the CPUs (and page-table walker) are in the same shareable domain (i.e innershareable). If you have a platform where it is not the case, then Xen is going to be badly broken.
Quoted 7 lines
This is also inline with the expectatio from the Arm Arm (B2-123 in DDI 0487D.a):
"The Inner Shareable domain is expected to be the set of PEs controlled by a single hypervisor or operating system."
Do you have a case where Xen needs to run on PEs in different domains?
No, thankfully I don't :-)
I was worried that some big.LITTLE SoCs might be built like that so. (I don't have any big.LITTLE machines here to confirm/deny.) It is good that we don't have to worry about it.
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 15/19 xen/arm: mm: Introduce DEFINE_PAGE_TABLE{, S} and use it
In reply to Julien GrallWe have multiple static page-tables defined in arch/arm/mm.c. The current way to define them is difficult to read and does not help when making modification.
Two new helpers DEFINE_PAGE_TABLES (to define multiple page-tables) and DEFINE_PAGE_TABLE (alias of DEFINE_PAGE_TABLES(..., 1)) are introduced and now used to define static page-tables.
Note that DEFINE_PAGE_TABLES() alignment differs from what is currently used for allocating page-tables. This is fine because page-tables are only required to be aligned to a page-size.
---
Changes in v2:
- Patch in replacement of "Use the shorter version __aligned(PAGE_SIZE) to align page-tables".
--- xen/arch/arm/mm.c | 32 ++++++++++++++++++-------------- 1 file changed, 18 insertions(+), 14 deletions(-)
xen/arch/arm/mm.c +18 -14
@@ -50,6 +50,11 @@ struct domain *dom_xen, *dom_io, *dom_cow;#undef mfn_to_virt#define mfn_to_virt (mfn) __mfn_to_virt(mfn_x(mfn)) +#define DEFINE_PAGE_TABLES (name, nr) \ +lpae_t __aligned(PAGE_SIZE) name[LPAE_ENTRIES * (nr)] + +#define DEFINE_PAGE_TABLE (name) DEFINE_PAGE_TABLES(name, 1) +/* Static start-of-day pagetables that we use before the allocators * are up. These are used by all CPUs during bringup before switching * to the CPUs own pagetables.
@@ -73,13 +78,13 @@ struct domain *dom_xen, *dom_io, *dom_cow;* Finally, if EARLY_PRINTK is enabled then xen_fixmap will be mapped * by the CPU once it has moved off the 1:1 mapping. */ -lpae_t boot_pgtable[LPAE_ENTRIES] __attribute__((__aligned__(4096))); +DEFINE_PAGE_TABLE (boot_pgtable);#ifdef CONFIG_ARM_64 -lpae_t boot_first[LPAE_ENTRIES] __attribute__((__aligned__(4096))); -lpae_t boot_first_id[LPAE_ENTRIES] __attribute__((__aligned__(4096))); +DEFINE_PAGE_TABLE (boot_first); +DEFINE_PAGE_TABLE (boot_first_id);#endif -lpae_t boot_second[LPAE_ENTRIES] __attribute__((__aligned__(4096))); -lpae_t boot_third[LPAE_ENTRIES] __attribute__((__aligned__(4096))); +DEFINE_PAGE_TABLE (boot_second); +DEFINE_PAGE_TABLE (boot_third);/* Main runtime page tables */
@@ -93,8 +98,8 @@ lpae_t boot_third[LPAE_ENTRIES] __attribute__((__aligned__(4096)));#ifdef CONFIG_ARM_64 #define HYP_PT_ROOT_LEVEL 0 -lpae_t xen_pgtable[LPAE_ENTRIES] __attribute__((__aligned__(4096))); -lpae_t xen_first[LPAE_ENTRIES] __attribute__((__aligned__(4096))); +static DEFINE_PAGE_TABLE (xen_pgtable); +static DEFINE_PAGE_TABLE (xen_first);#define THIS_CPU_PGTABLE xen_pgtable#else #define HYP_PT_ROOT_LEVEL 1
@@ -107,17 +112,16 @@ static DEFINE_PER_CPU(lpae_t *, xen_pgtable);* DOMHEAP_VIRT_START...DOMHEAP_VIRT_END in 2MB chunks. */ static DEFINE_PER_CPU(lpae_t *, xen_dommap);/* Root of the trie for cpu0, other CPU's PTs are dynamically allocated */ -lpae_t cpu0_pgtable[LPAE_ENTRIES] __attribute__((__aligned__(4096))); +static DEFINE_PAGE_TABLE (cpu0_pgtable);/* cpu0's domheap page tables */ -lpae_t cpu0_dommap[LPAE_ENTRIES*DOMHEAP_SECOND_PAGES] - __attribute__((__aligned__(4096*DOMHEAP_SECOND_PAGES))); +static DEFINE_PAGE_TABLES(cpu0_dommap, DOMHEAP_SECOND_PAGES);#endif #ifdef CONFIG_ARM_64 /* The first page of the first level mapping of the xenheap. The * subsequent xenheap first level pages are dynamically allocated, but * we need this one to bootstrap ourselves. */ -lpae_t xenheap_first_first[LPAE_ENTRIES] __attribute__((__aligned__(4096))); +static DEFINE_PAGE_TABLE (xenheap_first_first);/* The zeroeth level slot which uses xenheap_first_first. Used because * setup_xenheap_mappings otherwise relies on mfn_to_virt which isn't * valid for a non-xenheap mapping. */
@@ -131,12 +135,12 @@ static __initdata int xenheap_first_first_slot = -1;* addresses from 0 to 0x7fffffff. Offsets into it are calculated * with second_linear_offset(), not second_table_offset(). */ -lpae_t xen_second[LPAE_ENTRIES*2] __attribute__((__aligned__(4096*2))); +static DEFINE_PAGE_TABLES(xen_second, 2);/* First level page table used for fixmap */ -lpae_t xen_fixmap[LPAE_ENTRIES] __attribute__((__aligned__(4096))); +DEFINE_PAGE_TABLE (xen_fixmap);/* First level page table used to map Xen itself with the XN bit set * as appropriate. */ -static lpae_t xen_xenmap[LPAE_ENTRIES] __attribute__((__aligned__(4096))); +static DEFINE_PAGE_TABLE (xen_xenmap);/* Non-boot CPUs use this to find the correct pagetables. */ uint64_t init_ttbr;
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 15/19] xen/arm: mm: Introduce DEFINE_PAGE_TABLE{, S} and use it
Quoted 8 lines · Julien Grall
We have multiple static page-tables defined in
arch/arm/mm.c. The current way to define them is difficult to read and does not help when making modification.Two new helpers
DEFINE_PAGE_TABLES(to define multiple page-tables) andDEFINE_PAGE_TABLE(alias ofDEFINE_PAGE_TABLES(..., 1)) are introduced and now used to define static page-tables.
Quoted 5 lines
Note that
DEFINE_PAGE_TABLES()alignment differs from what is currently used for allocating page-tables. This is fine because page-tables are only required to be aligned to a page-size.Signed-off-by Julien Grall <[email protected]>
Quoted 94 lines
---
Changes in v2:
- Patch in replacement of "Use the shorter version__aligned(PAGE_SIZE)to align page-tables".--- xen/arch/arm/mm.c | 32 ++++++++++++++++++-------------- 1 file changed, 18 insertions(+), 14 deletions(-)xen/arch/arm/mm.c +18
-14@@ -50,6 +50,11 @@ struct domain *dom_xen, *dom_io, *dom_cow;#undef mfn_to_virt#define mfn_to_virt (mfn) __mfn_to_virt(mfn_x(mfn)) +#define DEFINE_PAGE_TABLES (name, nr) \ +lpae_t __aligned(PAGE_SIZE) name[LPAE_ENTRIES * (nr)] + +#define DEFINE_PAGE_TABLE (name) DEFINE_PAGE_TABLES(name, 1) +/* Static start-of-day pagetables that we use before the allocators * are up. These are used by all CPUs during bringup before switching * to the CPUs own pagetables. @@ -73,13 +78,13 @@ struct domain *dom_xen, *dom_io, *dom_cow;* Finally, if EARLY_PRINTK is enabled then xen_fixmap will be mapped * by the CPU once it has moved off the 1:1 mapping. */ -lpae_t boot_pgtable[LPAE_ENTRIES] __attribute__((__aligned__(4096))); +DEFINE_PAGE_TABLE (boot_pgtable);#ifdef CONFIG_ARM_64 -lpae_t boot_first[LPAE_ENTRIES] __attribute__((__aligned__(4096))); -lpae_t boot_first_id[LPAE_ENTRIES] __attribute__((__aligned__(4096))); +DEFINE_PAGE_TABLE (boot_first); +DEFINE_PAGE_TABLE (boot_first_id);#endif -lpae_t boot_second[LPAE_ENTRIES] __attribute__((__aligned__(4096))); -lpae_t boot_third[LPAE_ENTRIES] __attribute__((__aligned__(4096))); +DEFINE_PAGE_TABLE (boot_second); +DEFINE_PAGE_TABLE (boot_third);/* Main runtime page tables */ @@ -93,8 +98,8 @@ lpae_t boot_third[LPAE_ENTRIES] __attribute__((__aligned__(4096)));#ifdef CONFIG_ARM_64 #define HYP_PT_ROOT_LEVEL 0 -lpae_t xen_pgtable[LPAE_ENTRIES] __attribute__((__aligned__(4096))); -lpae_t xen_first[LPAE_ENTRIES] __attribute__((__aligned__(4096))); +static DEFINE_PAGE_TABLE (xen_pgtable); +static DEFINE_PAGE_TABLE (xen_first);#define THIS_CPU_PGTABLE xen_pgtable#else #define HYP_PT_ROOT_LEVEL 1@@ -107,17 +112,16 @@ static DEFINE_PER_CPU(lpae_t *, xen_pgtable);* DOMHEAP_VIRT_START...DOMHEAP_VIRT_END in 2MB chunks. */ static DEFINE_PER_CPU(lpae_t *, xen_dommap);/* Root of the trie for cpu0, other CPU's PTs are dynamically allocated */ -lpae_t cpu0_pgtable[LPAE_ENTRIES] __attribute__((__aligned__(4096))); +static DEFINE_PAGE_TABLE (cpu0_pgtable);/* cpu0's domheap page tables */ -lpae_t cpu0_dommap[LPAE_ENTRIES*DOMHEAP_SECOND_PAGES] - __attribute__((__aligned__(4096*DOMHEAP_SECOND_PAGES))); +static DEFINE_PAGE_TABLES(cpu0_dommap, DOMHEAP_SECOND_PAGES);#endif #ifdef CONFIG_ARM_64 /* The first page of the first level mapping of the xenheap. The * subsequent xenheap first level pages are dynamically allocated, but * we need this one to bootstrap ourselves. */ -lpae_t xenheap_first_first[LPAE_ENTRIES] __attribute__((__aligned__(4096))); +static DEFINE_PAGE_TABLE (xenheap_first_first);/* The zeroeth level slot which uses xenheap_first_first. Used because * setup_xenheap_mappings otherwise relies on mfn_to_virt which isn't * valid for a non-xenheap mapping. */ @@ -131,12 +135,12 @@ static __initdata int xenheap_first_first_slot = -1;* addresses from 0 to 0x7fffffff. Offsets into it are calculated * with second_linear_offset(), not second_table_offset(). */ -lpae_t xen_second[LPAE_ENTRIES*2] __attribute__((__aligned__(4096*2))); +static DEFINE_PAGE_TABLES(xen_second, 2);/* First level page table used for fixmap */ -lpae_t xen_fixmap[LPAE_ENTRIES] __attribute__((__aligned__(4096))); +DEFINE_PAGE_TABLE (xen_fixmap);/* First level page table used to map Xen itself with the XN bit set * as appropriate. */ -static lpae_t xen_xenmap[LPAE_ENTRIES] __attribute__((__aligned__(4096))); +static DEFINE_PAGE_TABLE (xen_xenmap);/* Non-boot CPUs use this to find the correct pagetables. */ uint64_t init_ttbr;Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 16/19 xen/arm: mm: Protect Xen page-table update with a spinlock
In reply to Julien GrallThe function create_xen_entries() may be called concurrently. For instance, while the vmap allocation is protected by a spinlock, the mapping is not.
The implementation create_xen_entries() contains quite a few TOCTOU races such as when allocating the 3rd-level page-tables.
Thankfully, they are pretty hard to reach as page-tables are allocated once and never released. Yet it is possible, so we need to protect with a spinlock to avoid corrupting the page-tables.
---
Changes in v2:
- Rework the commit message
--- xen/arch/arm/mm.c | 6 ++++++ 1 file changed, 6 insertions(+)
xen/arch/arm/mm.c +6 -0
@@ -974,6 +974,8 @@ enum xenmap_operation {RESERVE }; +static DEFINE_SPINLOCK (xen_pt_lock); + static int create_xen_entries(enum xenmap_operation op, unsigned long virt, mfn_t mfn,
@@ -985,6 +987,8 @@ static int create_xen_entries(enum xenmap_operation op,lpae_t pte, *entry;lpae_t *third =NULL ; +spin_lock (&xen_pt_lock); +for (; addr < addr_end; addr +=PAGE_SIZE , mfn =mfn_add (mfn,1 )) { entry = &xen_second[second_linear_offset (addr)];
@@ -1059,6 +1063,8 @@ out:*/ flush_xen_tlb_range_va (virt,PAGE_SIZE * nr_mfns); +spin_unlock (&xen_pt_lock); +return rc; }
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [PATCH MM-PART2 RESEND v2 16/19] xen/arm: mm: Protect Xen page-table update with a spinlock
Quoted 11 lines · Julien Grall
The function
create_xen_entries()may be called concurrently. For instance, while the vmap allocation is protected by a spinlock, the mapping is not.The implementation
create_xen_entries()contains quite a few TOCTOU races such as when allocating the 3rd-level page-tables.Thankfully, they are pretty hard to reach as page-tables are allocated once and never released. Yet it is possible, so we need to protect with a spinlock to avoid corrupting the page-tables.
Quoted 6 lines
Signed-off-by Julien Grall <[email protected]>---
Changes in v2:
- Rework the commit message
---
Signature · 7 lines
Sincerely, Andrii Anisov. _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 16/19] xen/arm: mm: Protect Xen page-table update with a spinlock
In reply to Julien GrallQuoted 3 lines · Julien Grall
The function
create_xen_entries()may be called concurrently. For instance, while the vmap allocation is protected by a spinlock, the mapping is not.
Do you have an example of potential concurrent calls of create_xen_entries() which doesn't involve concurrent vmaps (because vmaps are already protected by their spinlock)? vmap + something_else for instance?
Quoted 50 lines
The implementation
create_xen_entries()contains quite a few TOCTOU races such as when allocating the 3rd-level page-tables.Thankfully, they are pretty hard to reach as page-tables are allocated once and never released. Yet it is possible, so we need to protect with a spinlock to avoid corrupting the page-tables.
Signed-off-by Julien Grall <[email protected]>---
Changes in v2:
- Rework the commit message--- xen/arch/arm/mm.c | 6 ++++++ 1 file changed, 6 insertions(+)xen/arch/arm/mm.c +6
-0@@ -974,6 +974,8 @@ enum xenmap_operation {RESERVE }; +static DEFINE_SPINLOCK (xen_pt_lock); + static int create_xen_entries(enum xenmap_operation op, unsigned long virt, mfn_t mfn,@@ -985,6 +987,8 @@ static int create_xen_entries(enum xenmap_operation op,lpae_t pte, *entry;lpae_t *third =NULL ; +spin_lock (&xen_pt_lock); +for (; addr < addr_end; addr +=PAGE_SIZE , mfn =mfn_add (mfn,1 )) { entry = &xen_second[second_linear_offset (addr)];@@ -1059,6 +1063,8 @@ out:*/ flush_xen_tlb_range_va (virt,PAGE_SIZE * nr_mfns); +spin_unlock (&xen_pt_lock); +return rc; }Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 16/19] xen/arm: mm: Protect Xen page-table update with a spinlock
Hi Stefano,
Quoted 9 lines · Stefano Stabellini
Quoted 3 lines · Julien Grall
The function
create_xen_entries()may be called concurrently. For instance, while the vmap allocation is protected by a spinlock, the mapping is not.Do you have an example of potential concurrent calls of
create_xen_entries()which doesn't involve concurrent vmaps (because vmaps are already protected by their spinlock)? vmap +something_elsefor instance?
Well, I gave an example here. The vmap allocation (i.e vm_alloc) is protected by a spinlock. However, when the mapping is done there are no spinlock to protected against concurrent one.
So the following scenario could happen:
CPU0 | CPU1
|
vmap() | vmap()
vm_alloc() | vm_alloc()
spin_lock() |
... |
spin_unlock() |
| spin_lock()
* interrupt * | ...
| spin_unlock()
|
map_pages_to_xen() | map_pages_to_xen()
entry = &xen_second[X] | entry = &xen_second[Y]
* entry invalid * | * entry invalid *
create_xen_table() | create_xen_table()Assuming X == Y (i.e we the xen second entry is the same), then one will win the race and therefore one mapping will be inexistent.
But as I wrote, the chance it is happening is very limited.
I can add that in the commit message.
Cheers,
Signature · 6 lines
Julien Grall _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Stefano Stabellini
Reviewed-byRe: [Xen-devel] [PATCH MM-PART2 RESEND v2 16/19] xen/arm: mm: Protect Xen page-table update with a spinlock
Quoted 35 lines · Julien Grall
Hi Stefano,
Quoted 9 lines · Stefano Stabellini
Quoted 3 lines · Julien Grall
The function
create_xen_entries()may be called concurrently. For instance, while the vmap allocation is protected by a spinlock, the mapping is not.Do you have an example of potential concurrent calls of
create_xen_entries()which doesn't involve concurrent vmaps (because vmaps are already protected by their spinlock)? vmap +something_elsefor instance?Well, I gave an example here. The vmap allocation (i.e
vm_alloc) is protected by a spinlock. However, when the mapping is done there are no spinlock to protected against concurrent one.So the following scenario could happen:
CPU0 | CPU1
| vmap() | vmap() vm_alloc() | vm_alloc() spin_lock() | ... | spin_unlock() | | spin_lock() * interrupt * | ... | spin_unlock() | map_pages_to_xen() | map_pages_to_xen() entry = &xen_second[X] | entry = &xen_second[Y] * entry invalid * | * entry invalid * create_xen_table() | create_xen_table()
Quoted 6 lines
Assuming X == Y (i.e we the xen second entry is the same), then one will win the race and therefore one mapping will be inexistent.
But as I wrote, the chance it is happening is very limited.
I can add that in the commit message.
Thanks for the detailed explanation, I am just trying to understand and double-check the race. Wouldn't vm_alloc guarantee to return a different va in the two cases (CPU0 and CPU1 above), given that the selection of the va is done under spin_lock? But it would be still possible to have X and Y so that they select the same &xen_second entry, hence, the race with create_xen_table(). It looks like the race is there.
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 16/19] xen/arm: mm: Protect Xen page-table update with a spinlock
Hi Stefano,
Quoted 37 lines · Stefano Stabellini
Quoted 36 lines · Julien Grall
Quoted 9 lines · Stefano Stabellini
Quoted 3 lines · Julien Grall
The function
create_xen_entries()may be called concurrently. For instance, while the vmap allocation is protected by a spinlock, the mapping is not.Do you have an example of potential concurrent calls of
create_xen_entries()which doesn't involve concurrent vmaps (because vmaps are already protected by their spinlock)? vmap +something_elsefor instance?Well, I gave an example here. The vmap allocation (i.e
vm_alloc) is protected by a spinlock. However, when the mapping is done there are no spinlock to protected against concurrent one.So the following scenario could happen:
CPU0 | CPU1
| vmap() | vmap() vm_alloc() | vm_alloc() spin_lock() | ... | spin_unlock() | | spin_lock() * interrupt * | ... | spin_unlock() | map_pages_to_xen() | map_pages_to_xen() entry = &xen_second[X] | entry = &xen_second[Y] * entry invalid * | * entry invalid * create_xen_table() | create_xen_table()Assuming X == Y (i.e we the xen second entry is the same), then one will win the race and therefore one mapping will be inexistent.
Quoted 8 lines
Quoted 3 lines
But as I wrote, the chance it is happening is very limited.
I can add that in the commit message.
Thanks for the detailed explanation, I am just trying to understand and double-check the race. Wouldn't
vm_allocguarantee to return a different va in the two cases (CPU0 and CPU1 above), given that the selection of the va is done underspin_lock?
Yes vm_alloc() would guarantee you to have a different VA.
Quoted 3 lines
But it would be still possible to have X
and Y so that they select the same&xen_secondentry, hence, the race withcreate_xen_table(). It looks like the race is there.
That's correct.
Quoted 2 lines
Reviewed-by Stefano Stabellini <[email protected]>
Thank you!
Cheers,
Signature · 6 lines
Julien Grall _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 17/19 xen/arm: mm: Initialize page-tables earlier
In reply to Julien GrallSince commit f60658c6ae "xen/arm: Stop relocating Xen", the function setup_page_tables() does not require any information from the FDT.
So the initialization of the page-tables can be done much earlier in the boot process. The earliest setup_page_tables() can be called is after traps have been initialized, so we can get backtrace if an error occurred.
Moving the initialization of the page-tables also avoid the dance to map the FDT again in the new set of page-tables.
---
Changes in v2:
- Add Andrii's reviewed-by
--- xen/arch/arm/mm.c | 12 +++--------- xen/arch/arm/setup.c | 4 ++-- 2 files changed, 5 insertions(+), 11 deletions(-)
xen/arch/arm/mm.c +3 -9
@@ -550,7 +550,7 @@ static inline lpae_t pte_of_xenaddr(vaddr_t va)return mfn_to_xen_entry (maddr_to_mfn (ma),MT_NORMAL ); } -/* Map the FDT in the early boot page table */ +/* Map the FDT in the runtime page table */ void *__init early_fdt_map (paddr_t fdt_paddr) {/* We are using 2MB superpage for mapping the FDT */
@@ -573,7 +573,7 @@ void * __init early_fdt_map(paddr_t fdt_paddr)/* The FDT is mapped using 2MB superpage */ BUILD_BUG_ON (BOOT_FDT_VIRT_START %SZ_2M ); -create_mappings (boot_second,BOOT_FDT_VIRT_START ,paddr_to_pfn (base_paddr), +create_mappings (xen_second,BOOT_FDT_VIRT_START ,paddr_to_pfn (base_paddr),SZ_2M >>PAGE_SHIFT ,SZ_2M ); offset = fdt_paddr %SECOND_SIZE ;
@@ -588,7 +588,7 @@ void * __init early_fdt_map(paddr_t fdt_paddr)if ( (offset + size) >SZ_2M ) { -create_mappings (boot_second,BOOT_FDT_VIRT_START +SZ_2M , +create_mappings (xen_second,BOOT_FDT_VIRT_START +SZ_2M ,paddr_to_pfn (base_paddr +SZ_2M ),SZ_2M >>PAGE_SHIFT ,SZ_2M ); }
@@ -699,12 +699,6 @@ void __init setup_pagetables(unsigned long boot_phys_offset) pte.pt.table =1 ; xen_second[second_table_offset (FIXMAP_ADDR (0 ))] = pte; -/* ... DTB */ - pte = boot_second[second_table_offset (BOOT_FDT_VIRT_START )]; - xen_second[second_table_offset (BOOT_FDT_VIRT_START )] = pte; - pte = boot_second[second_table_offset (BOOT_FDT_VIRT_START +SZ_2M )]; - xen_second[second_table_offset (BOOT_FDT_VIRT_START +SZ_2M )] = pte; -#ifdef CONFIG_ARM_64 ttbr = (uintptr_t ) xen_pgtable + phys_offset;#else
xen/arch/arm/setup.c +2 -2
@@ -759,6 +759,8 @@ void __init start_xen(unsigned long boot_phys_offset,/* Initialize traps early allow us to get backtrace when an error occurred */ init_traps (); +setup_pagetables (boot_phys_offset); +smp_clear_cpu_maps (); device_tree_flattened =early_fdt_map (fdt_paddr);
@@ -780,8 +782,6 @@ void __init start_xen(unsigned long boot_phys_offset, (paddr_t)(uintptr_t)(_end - _start + 1), false);BUG_ON (!xen_bootmodule); -setup_pagetables (boot_phys_offset); -setup_mm (fdt_paddr, fdt_size);/* Parse the ACPI tables for possible boot-time configuration */
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 17/19] xen/arm: mm: Initialize page-tables earlier
Quoted 27 lines · Julien Grall
Since commit
f60658c6ae"xen/arm: Stop relocating Xen", the functionsetup_page_tables()does not require any information from the FDT.So the initialization of the page-tables can be done much earlier in the boot process. The earliest
setup_page_tables()can be called is after traps have been initialized, so we can get backtrace if an error occurred.Moving the initialization of the page-tables also avoid the dance to map the FDT again in the new set of page-tables.
Signed-off-by Julien Grall <[email protected]>Reviewed-by Andrii Anisov <[email protected]>---
Changes in v2:
- Add Andrii's reviewed-by--- xen/arch/arm/mm.c | 12 +++--------- xen/arch/arm/setup.c | 4 ++-- 2 files changed, 5 insertions(+), 11 deletions(-)xen/arch/arm/mm.c +0
-0
xen/arch/arm/mm.c static inline lpae_t pte_of_xenaddr(vaddr_t va) · line 553
Quoted 5 lines
xen/arch/arm/mm.c +1
-1@@ -550,7 +550,7 @@ static inline lpae_t pte_of_xenaddr(vaddr_t va)return mfn_to_xen_entry (maddr_to_mfn (ma),MT_NORMAL ); } -/* Map the FDT in the early boot page table */ +/* Map the FDT in the runtime page table */
I think you can drop this comment now.
In any case:
Quoted 59 lines
xen/arch/arm/mm.c +2
-8void *__init early_fdt_map (paddr_t fdt_paddr) {/* We are using 2MB superpage for mapping the FDT */ @@ -573,7 +573,7 @@ void * __init early_fdt_map(paddr_t fdt_paddr)/* The FDT is mapped using 2MB superpage */ BUILD_BUG_ON (BOOT_FDT_VIRT_START %SZ_2M ); -create_mappings (boot_second,BOOT_FDT_VIRT_START ,paddr_to_pfn (base_paddr), +create_mappings (xen_second,BOOT_FDT_VIRT_START ,paddr_to_pfn (base_paddr),SZ_2M >>PAGE_SHIFT ,SZ_2M ); offset = fdt_paddr %SECOND_SIZE ;@@ -588,7 +588,7 @@ void * __init early_fdt_map(paddr_t fdt_paddr)if ( (offset + size) >SZ_2M ) { -create_mappings (boot_second,BOOT_FDT_VIRT_START +SZ_2M , +create_mappings (xen_second,BOOT_FDT_VIRT_START +SZ_2M ,paddr_to_pfn (base_paddr +SZ_2M ),SZ_2M >>PAGE_SHIFT ,SZ_2M ); }@@ -699,12 +699,6 @@ void __init setup_pagetables(unsigned long boot_phys_offset) pte.pt.table =1 ; xen_second[second_table_offset (FIXMAP_ADDR (0 ))] = pte; -/* ... DTB */ - pte = boot_second[second_table_offset (BOOT_FDT_VIRT_START )]; - xen_second[second_table_offset (BOOT_FDT_VIRT_START )] = pte; - pte = boot_second[second_table_offset (BOOT_FDT_VIRT_START +SZ_2M )]; - xen_second[second_table_offset (BOOT_FDT_VIRT_START +SZ_2M )] = pte; -#ifdef CONFIG_ARM_64 ttbr = (uintptr_t ) xen_pgtable + phys_offset;#else xen/arch/arm/setup.c +2
-2@@ -759,6 +759,8 @@ void __init start_xen(unsigned long boot_phys_offset,/* Initialize traps early allow us to get backtrace when an error occurred */ init_traps (); +setup_pagetables (boot_phys_offset); +smp_clear_cpu_maps (); device_tree_flattened =early_fdt_map (fdt_paddr);@@ -780,8 +782,6 @@ void __init start_xen(unsigned long boot_phys_offset, (paddr_t)(uintptr_t)(_end - _start + 1), false);BUG_ON (!xen_bootmodule); -setup_pagetables (boot_phys_offset); -setup_mm (fdt_paddr, fdt_size);/* Parse the ACPI tables for possible boot-time configuration */ Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 17/19] xen/arm: mm: Initialize page-tables earlier
Hi Stefano,
Quoted 31 lines · Stefano Stabellini
Quoted 30 lines · Julien Grall
Since commit
f60658c6ae"xen/arm: Stop relocating Xen", the functionsetup_page_tables()does not require any information from the FDT.So the initialization of the page-tables can be done much earlier in the boot process. The earliest
setup_page_tables()can be called is after traps have been initialized, so we can get backtrace if an error occurred.Moving the initialization of the page-tables also avoid the dance to map the FDT again in the new set of page-tables.
Signed-off-by Julien Grall <[email protected]>Reviewed-by Andrii Anisov <[email protected]>---
Changes in v2:
- Add Andrii's reviewed-by--- xen/arch/arm/mm.c | 12 +++--------- xen/arch/arm/setup.c | 4 ++-- 2 files changed, 5 insertions(+), 11 deletions(-)xen/arch/arm/mm.c +0
-0@@ -550,7 +550,7 @@ static inline lpae_t pte_of_xenaddr(vaddr_t va)return mfn_to_xen_entry (maddr_to_mfn (ma),MT_NORMAL ); }
Quoted 4 lines
xen/arch/arm/mm.c static inline lpae_t pte_of_xenaddr(vaddr_t va) · line 553
Quoted 2 lines
xen/arch/arm/mm.c +1
-1-/* Map the FDT in the early boot page table */ +/* Map the FDT in the runtime page table */ I think you can drop this comment now.
Good point, the more the plan is to have only one set of page-tables.
Quoted 4 lines
In any case:
Reviewed-by Stefano Stabellini <[email protected]>
Thank you for the review!
Cheers,
Signature · 6 lines
Julien Grall _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 18/19 xen/arm: mm: Check start is always before end in {destroy, modify}_xen_mappings
In reply to Julien GrallThe two helpers {destroy, modify}_xen_mappings don't check that the start is always before the end. This should never happen but if it happens, it will result to unexpected behavior.
Catch such issues earlier on by adding an ASSERT in destroy_xen_mappings and modify_xen_mappings.
---
Changes in v2:
- Add Andrii's reviewed-by
--- xen/arch/arm/mm.c | 2 ++ 1 file changed, 2 insertions(+)
xen/arch/arm/mm.c +2 -0
@@ -1077,11 +1077,13 @@ int populate_pt_range(unsigned long virt, unsigned long nr_mfns)int destroy_xen_mappings (unsigned long v,unsigned long e) { +ASSERT (v <= e);return create_xen_entries (REMOVE , v,INVALID_MFN , (e - v) >>PAGE_SHIFT ,0 ); }int modify_xen_mappings (unsigned long s,unsigned long e,unsigned int flags) { +ASSERT (s <= e);return create_xen_entries (MODIFY , s,INVALID_MFN , (e - s) >>PAGE_SHIFT , flags); }
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 18/19] xen/arm: mm: Check start is always before end in {destroy, modify}_xen_mappings
Quoted 9 lines · Julien Grall
The two helpers {destroy, modify}_xen_mappings don't check that the start is always before the end. This should never happen but if it happens, it will result to unexpected behavior.
Catch such issues earlier on by adding an ASSERT in
destroy_xen_mappingsandmodify_xen_mappings.Signed-off-by Julien Grall <[email protected]>Reviewed-by Andrii Anisov <[email protected]>
Quoted 28 lines
---
Changes in v2:
- Add Andrii's reviewed-by--- xen/arch/arm/mm.c | 2 ++ 1 file changed, 2 insertions(+)xen/arch/arm/mm.c +2
-0@@ -1077,11 +1077,13 @@ int populate_pt_range(unsigned long virt, unsigned long nr_mfns)int destroy_xen_mappings (unsigned long v,unsigned long e) { +ASSERT (v <= e);return create_xen_entries (REMOVE , v,INVALID_MFN , (e - v) >>PAGE_SHIFT ,0 ); }int modify_xen_mappings (unsigned long s,unsigned long e,unsigned int flags) { +ASSERT (s <= e);return create_xen_entries (MODIFY , s,INVALID_MFN , (e - s) >>PAGE_SHIFT , flags); }Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
PATCH RESEND v2 19/19 xen/arm: Pair call to set_fixmap with call to clear_fixmap in copy_from_paddr
In reply to Julien GrallAt the moment, set_fixmap may replace a valid entry without following the break-before-make sequence. This may result to TLB conflict abort.
Rather than dealing with Break-Before-Make in set_fixmap, every call to set_fixmap is paired with a call to clear_fixmap.
---
Changes in v2:
- Add Andrii's reviewed-by
--- xen/arch/arm/kernel.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)
xen/arch/arm/kernel.c +1 -2
@@ -58,13 +58,12 @@ void __init copy_from_paddr(void *dst, paddr_t paddr, unsigned long len)set_fixmap (FIXMAP_MISC ,maddr_to_mfn (paddr),PAGE_HYPERVISOR_WC );memcpy (dst, src + s, l);clean_dcache_va_range (dst, l); +clear_fixmap (FIXMAP_MISC ); paddr += l; dst += l; len -= l; } - -clear_fixmap (FIXMAP_MISC ); } static void __init place_modules(struct kernel_info *info,
Signature · 7 lines
2.11.0 _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 19/19] xen/arm: Pair call to set_fixmap with call to clear_fixmap in copy_from_paddr
Quoted 5 lines · Julien Grall
At the moment,
set_fixmapmay replace a valid entry without following the break-before-make sequence. This may result to TLB conflict abort.Rather than dealing with Break-Before-Make in
set_fixmap, every call toset_fixmapis paired with a call toclear_fixmap.
It is not every call to set_fixmap: it is every call to set_fixmap(FIXMAP_MISC, ...
Please clarify, then you can add
Quoted 32 lines
Signed-off-by Julien Grall <[email protected]>Reviewed-by Andrii Anisov <[email protected]>---
Changes in v2:
- Add Andrii's reviewed-by--- xen/arch/arm/kernel.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-)xen/arch/arm/kernel.c +1
-2@@ -58,13 +58,12 @@ void __init copy_from_paddr(void *dst, paddr_t paddr, unsigned long len)set_fixmap (FIXMAP_MISC ,maddr_to_mfn (paddr),PAGE_HYPERVISOR_WC );memcpy (dst, src + s, l);clean_dcache_va_range (dst, l); +clear_fixmap (FIXMAP_MISC ); paddr += l; dst += l; len -= l; } - -clear_fixmap (FIXMAP_MISC ); } static void __init place_modules(struct kernel_info *info,Signature · 1 line
2.11.0
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 19/19] xen/arm: Pair call to set_fixmap with call to clear_fixmap in copy_from_paddr
Quoted 9 lines · Stefano Stabellini
Quoted 5 lines · Julien Grall
At the moment,
set_fixmapmay replace a valid entry without following the break-before-make sequence. This may result to TLB conflict abort.Rather than dealing with Break-Before-Make in
set_fixmap, every call toset_fixmapis paired with a call toclear_fixmap.It is not every call to
set_fixmap: it is every call to set_fixmap(FIXMAP_MISC, ...
I don't understand this request... The title explicit mention "copy_from_paddr" and fixmap is only called with FIXMAP_MISC.
So why should I need to specify the argument?
Quoted 4 lines
Please clarify, then you can add
Reviewed-by Stefano Stabellini <[email protected]>
Cheers,
Signature · 6 lines
Julien Grall _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 19/19] xen/arm: Pair call to set_fixmap with call to clear_fixmap in copy_from_paddr
Quoted 8 lines · Julien Grall
Quoted 7 lines · Stefano Stabellini
Quoted 5 lines · Julien Grall
At the moment,
set_fixmapmay replace a valid entry without following the break-before-make sequence. This may result to TLB conflict abort.Rather than dealing with Break-Before-Make in
set_fixmap, every call toset_fixmapis paired with a call toclear_fixmap.
Quoted 7 lines
Quoted 2 lines
It is not every call to
set_fixmap: it is every call to set_fixmap(FIXMAP_MISC, ...I don't understand this request... The title explicit mention "
copy_from_paddr" and fixmap is only called withFIXMAP_MISC.So why should I need to specify the argument?
I wasn't asking to mention FIXMAP_MISC explicitely, I don't think it is particularly useful. I was only trying to make the wording more specific to what the patch does.
The statement "every call to set_fixmap is paired with a call to clear_fixmap" is too generic and I would prefer if it was limited in scope by something like
"in copy_from_paddr"
Like you have done in the subject. Resulting in:
every call to set_fixmap in copy_from_paddr is paired with a call to clear_fixmap
Quoted 3 lines
Quoted 3 lines
Please clarify, then you can add
Reviewed-by Stefano Stabellini <[email protected]>
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
Re: [Xen-devel] [PATCH MM-PART2 RESEND v2 19/19] xen/arm: Pair call to set_fixmap with call to clear_fixmap in copy_from_paddr
Hi Stefano,
Quoted 12 lines · Stefano Stabellini
Quoted 11 lines · Julien Grall
Quoted 9 lines · Stefano Stabellini
Quoted 5 lines · Julien Grall
At the moment,
set_fixmapmay replace a valid entry without following the break-before-make sequence. This may result to TLB conflict abort.Rather than dealing with Break-Before-Make in
set_fixmap, every call toset_fixmapis paired with a call toclear_fixmap.It is not every call to
set_fixmap: it is every call to set_fixmap(FIXMAP_MISC, ...
Quoted 8 lines
Quoted 4 lines
I don't understand this request... The title explicit mention "
copy_from_paddr" and fixmap is only called withFIXMAP_MISC.So why should I need to specify the argument?
I wasn't asking to mention
FIXMAP_MISCexplicitely, I don't think it is particularly useful. I was only trying to make the wording more specific to what the patch does.
I have to admit this wasn't clear from your answer. Anyway,...
Quoted 11 lines
The statement "every call to
set_fixmapis paired with a call toclear_fixmap" is too generic and I would prefer if it was limited in scope by something like"in
copy_from_paddr"Like you have done in the subject. Resulting in:
every call to
set_fixmapincopy_from_paddris paired with a call toclear_fixmap
... thank you for your clarification. I have updated the commit message accordingly.
Cheers,
Signature · 6 lines
Julien Grall _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Hi,
I am missing some reply/review from Stefano (see below).
Quoted 2 lines · Julien Grall
Julien Grall (19):
xen/arm: ReworkHSCTLR_BASE
More details input on his claim... and review for the following patches:
Quoted 5 lines
xen/arm: Remove parameter cpuid from
start_xenxen/arm32: head: Correctly report the HW CPU ID xen/arm32: head: Don't set MAIR0 and MAIR1
xen/arm32: head: Always zero r3 before update a page-table entry xen/arm32: mm: Avoid to zero and clean cache for CPU0 domheap
Quoted 8 lines
xen/arm32: mm: Avoid cleaning the cache for secondary CPUs page-tables xen/arm: mm: Introduce DEFINE_PAGE_TABLE{,S} and use it xen/arm: mm: Protect Xen page-table update with a spinlock xen/arm: mm: Initialize page-tables earlier
xen/arm: mm: Check start is always before end in {destroy,
modify}_xen_mappings
xen/arm: Pair call toset_fixmapwith call toclear_fixmapin
copy_from_paddr
Cheers,
Signature · 6 lines
Julien Grall _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
Hi,
Quoted 8 lines · Julien Grall
This is the second part of the boot/memory rework for Xen on Arm. This part contains mostly clean-up & fixes found during the rework.
The first part of the rework is "xen/arm: TLB flush helpers rework" [1].
For convenience, I provided a branch with all the patches applied based on staging:
Quoted 8 lines
git://xenbits.xen.org/people/julieng/xen-unstable.gitbranchmm/part2/v2Cheers,
[1] https://lists.xenproject.org/archives/html/xen-devel/2019-05/msg01109.html
Julien Grall (19):
xen/arm: ReworkHSCTLR_BASE
I have committed all but this patch as it needs a respin. I will send it as part of MM-PART3.
Thank you for the review.
Cheers,
Signature · 6 lines
Julien Grall _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel