Page MenuHomeFreeBSD
Differential D35922

stand: Use devformat instead of disk_devfmt
ClosedPublic
Actions

Authored by imp on Jul 25 2022, 8:33 PM.
Tags
None
Referenced Files
Unknown Object (File)
Tue, Mar 11, 11:24 PM
Unknown Object (File)
Wed, Feb 26, 10:23 PM
Unknown Object (File)
Wed, Feb 26, 8:17 PM
Unknown Object (File)
Tue, Feb 25, 11:18 PM
Unknown Object (File)
Tue, Feb 25, 10:48 PM
Unknown Object (File)
Tue, Feb 25, 12:23 PM
Unknown Object (File)
Jan 6 2025, 7:40 AM
Unknown Object (File)
Dec 22 2024, 3:18 AM
View All 80 Files
Subscribers
dab

Details

Reviewers
tsoome
kevans
manu
Commits
rGec6d9023e56a: stand: Use devformat instead of disk_devfmt
rG0b3a4a588fb6: stand: Use devformat instead of disk_devfmt
Summary

Use devformat instead of disk_devfmt. This allows us to avoid knowing
the details of the device that's underneath us. Remove disk.h include
and the -I${LDRSRC} from the build of ufs.c since they are no longer
needed.

Sponsored by: Netflix

Diff Detail

Repository
rS FreeBSD src repository - subversion
Lint
Lint Passed
Unit
No Test Coverage
Build Status
Buildable 46652
Build 43541: arc lint + arc unit

Revision Contents
Changeset List

PathSize
stand/
libsa/
1 line
5 lines

Diff 108659

View Options

stand/libsa/Makefile

Show First 20 Lines • Show All 194 Lines • ▼ Show 20 Lines/* SACK related state */
struct osd *osd; /* storage for Khelp module data */ struct osd *osd; /* storage for Khelp module data */
u_int t_keepinit; /* time to establish connection */ u_int t_keepinit; /* time to establish connection */
u_int t_keepidle; /* time before keepalive probes begin */ u_int t_keepidle; /* time before keepalive probes begin */
u_int t_keepintvl; /* interval between keepalives */ u_int t_keepintvl; /* interval between keepalives */
u_int t_keepcnt; /* number of keepalives before close */ u_int t_keepcnt; /* number of keepalives before close */
u_int t_tsomax; /* tso burst length limit */ u_int t_tsomax; /* tso burst length limit */
u_int t_pmtud_saved_maxopd; /* pre-blackhole MSS */
hirenUnsubmitted
Not Done
Inline Actions

indentation issue?

hiren: indentation issue?
u_int t_pmtud_flags; /* PLPMTU flags */
hirenUnsubmitted
Not Done
Inline Actions

I do not think naming this flag specific to pmtud is a good idea unless you'll use most of the bits available in PMTUD/PLPMTUD implementation.

IMO, we should give this a bit more generic name so others can use remaining bits in future?

hiren: I do not think naming this flag specific to pmtud is a good idea unless you'll use most of the…
sbrunoAuthorUnsubmitted
Not Done
Inline Actions

I actually think I might be using a large amount of bits here for flags and indicators when the actual PLPMTUD specification is implemented this year. I can already see 5 or so more flags being allocated which probably means I'm going to need 2x of that. That's not 32bits, but that's still quite a lot for one feature and leaves room for additional things in the future.

sbruno: I actually think I might be using a large amount of bits here for flags and indicators when the…
bzUnsubmitted
Not Done
Inline Actions

I wonder if this should be t_flags2 as t_flags will run out after one more bit and we only need 3.

bz: I wonder if this should be t_flags2 as t_flags will run out after one more bit and we only need…
uint32_t t_ispare[8]; /* 5 UTO, 3 TBD */ uint32_t t_ispare[8]; /* 5 UTO, 3 TBD */
hirenUnsubmitted
Not Done
Inline Actions

I think the comment should be "5 UTO, 2 TBD".

hiren: I think the comment should be "5 UTO, 2 TBD".
bzUnsubmitted
Not Done
Inline Actions

You do not use (unassigned) spares on HEAD. Keep the fields next to the unassigned spares however and you have a good time merging without breaking things.

bz: You do not use (unassigned) spares on HEAD. Keep the fields next to the unassigned spares…
sbrunoAuthorUnsubmitted
Not Done
Inline Actions

To be clear, for HEAD just add elements to the struct. Don't consume spares.

For MFC to stable releases, consume spares, don't add elements to the struct.

Is this what you are trying to tell me?

sbruno: To be clear, for HEAD just add elements to the struct. Don't consume spares. For MFC to…
bzUnsubmitted
Not Done
Inline Actions

You are still decrementing "TBD" spares on head, which you should not.

bz: You are still decrementing "TBD" spares on head, which you should not.
void *t_pspare2[4]; /* 1 TCP_SIGNATURE, 3 TBD */ void *t_pspare2[4]; /* 1 TCP_SIGNATURE, 3 TBD */
uint64_t _pad[6]; /* 6 TBD (1-2 CC/RTT?) */ uint64_t _pad[6]; /* 6 TBD (1-2 CC/RTT?) */
}; };
/* /*
* Flags and utility macros for the t_flags field. * Flags and utility macros for the t_flags field.
*/ */
#define TF_ACKNOW 0x000001 /* ack peer immediately */ #define TF_ACKNOW 0x000001 /* ack peer immediately */
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
#define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */ #define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */
#define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */ #define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */
/* /*
* Only a single SA per host may be specified at this time. An SPI is * Only a single SA per host may be specified at this time. An SPI is
* needed in order for the KEY_ALLOCSA() lookup to work. * needed in order for the KEY_ALLOCSA() lookup to work.
*/ */
#define TCP_SIG_SPI 0x1000 #define TCP_SIG_SPI 0x1000
#endif /* TCP_SIGNATURE */ #endif /* TCP_SIGNATURE */
/*
* Flags for PLPMTU handling, t_pmtu_flags
*/
#define PLPMTU_BLACKHOLE 0x00000001 /* possible PLPMTUD Black Hole */
#define PLPMTU_PMTUD 0x00000002 /* allowed to attempt PLPMTUD */
#define PLPMTU_MAXSEGSNT 0x00000004 /* last segment sent was full seg */
bzUnsubmitted
Not Done
Inline Actions

#define<tab>FOO<tab>BAR<space>/* Blah. */ will work best.

bz: #define<tab>FOO<tab>BAR<space>/* Blah. */ will work best.
/* /*
* Structure to hold TCP options that are only used during segment * Structure to hold TCP options that are only used during segment
* processing (in tcp_input), but not held in the tcpcb. * processing (in tcp_input), but not held in the tcpcb.
* It's basically used to reduce the number of parameters * It's basically used to reduce the number of parameters
* to tcp_dooptions and tcp_addoptions. * to tcp_dooptions and tcp_addoptions.
* The binary order of the to_flags is relevant for packing of the * The binary order of the to_flags is relevant for packing of the
* options in tcp_addoptions. * options in tcp_addoptions.
*/ */
▲ Show 20 Lines • Show All 460 Lines • Show Last 20 Lines
View Options

stand/libsa/ufs.c

Show First 20 Lines • Show All 194 Lines • ▼ Show 20 Lines/* SACK related state */
struct osd *osd; /* storage for Khelp module data */ struct osd *osd; /* storage for Khelp module data */
u_int t_keepinit; /* time to establish connection */ u_int t_keepinit; /* time to establish connection */
u_int t_keepidle; /* time before keepalive probes begin */ u_int t_keepidle; /* time before keepalive probes begin */
u_int t_keepintvl; /* interval between keepalives */ u_int t_keepintvl; /* interval between keepalives */
u_int t_keepcnt; /* number of keepalives before close */ u_int t_keepcnt; /* number of keepalives before close */
u_int t_tsomax; /* tso burst length limit */ u_int t_tsomax; /* tso burst length limit */
u_int t_pmtud_saved_maxopd; /* pre-blackhole MSS */
hirenUnsubmitted
Not Done
Inline Actions

indentation issue?

hiren: indentation issue?
u_int t_pmtud_flags; /* PLPMTU flags */
hirenUnsubmitted
Not Done
Inline Actions

I do not think naming this flag specific to pmtud is a good idea unless you'll use most of the bits available in PMTUD/PLPMTUD implementation.

IMO, we should give this a bit more generic name so others can use remaining bits in future?

hiren: I do not think naming this flag specific to pmtud is a good idea unless you'll use most of the…
sbrunoAuthorUnsubmitted
Not Done
Inline Actions

I actually think I might be using a large amount of bits here for flags and indicators when the actual PLPMTUD specification is implemented this year. I can already see 5 or so more flags being allocated which probably means I'm going to need 2x of that. That's not 32bits, but that's still quite a lot for one feature and leaves room for additional things in the future.

sbruno: I actually think I might be using a large amount of bits here for flags and indicators when the…
bzUnsubmitted
Not Done
Inline Actions

I wonder if this should be t_flags2 as t_flags will run out after one more bit and we only need 3.

bz: I wonder if this should be t_flags2 as t_flags will run out after one more bit and we only need…
uint32_t t_ispare[8]; /* 5 UTO, 3 TBD */ uint32_t t_ispare[8]; /* 5 UTO, 3 TBD */
hirenUnsubmitted
Not Done
Inline Actions

I think the comment should be "5 UTO, 2 TBD".

hiren: I think the comment should be "5 UTO, 2 TBD".
bzUnsubmitted
Not Done
Inline Actions

You do not use (unassigned) spares on HEAD. Keep the fields next to the unassigned spares however and you have a good time merging without breaking things.

bz: You do not use (unassigned) spares on HEAD. Keep the fields next to the unassigned spares…
sbrunoAuthorUnsubmitted
Not Done
Inline Actions

To be clear, for HEAD just add elements to the struct. Don't consume spares.

For MFC to stable releases, consume spares, don't add elements to the struct.

Is this what you are trying to tell me?

sbruno: To be clear, for HEAD just add elements to the struct. Don't consume spares. For MFC to…
bzUnsubmitted
Not Done
Inline Actions

You are still decrementing "TBD" spares on head, which you should not.

bz: You are still decrementing "TBD" spares on head, which you should not.
void *t_pspare2[4]; /* 1 TCP_SIGNATURE, 3 TBD */ void *t_pspare2[4]; /* 1 TCP_SIGNATURE, 3 TBD */
uint64_t _pad[6]; /* 6 TBD (1-2 CC/RTT?) */ uint64_t _pad[6]; /* 6 TBD (1-2 CC/RTT?) */
}; };
/* /*
* Flags and utility macros for the t_flags field. * Flags and utility macros for the t_flags field.
*/ */
#define TF_ACKNOW 0x000001 /* ack peer immediately */ #define TF_ACKNOW 0x000001 /* ack peer immediately */
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
#define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */ #define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */
#define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */ #define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */
/* /*
* Only a single SA per host may be specified at this time. An SPI is * Only a single SA per host may be specified at this time. An SPI is
* needed in order for the KEY_ALLOCSA() lookup to work. * needed in order for the KEY_ALLOCSA() lookup to work.
*/ */
#define TCP_SIG_SPI 0x1000 #define TCP_SIG_SPI 0x1000
#endif /* TCP_SIGNATURE */ #endif /* TCP_SIGNATURE */
/*
* Flags for PLPMTU handling, t_pmtu_flags
*/
#define PLPMTU_BLACKHOLE 0x00000001 /* possible PLPMTUD Black Hole */
#define PLPMTU_PMTUD 0x00000002 /* allowed to attempt PLPMTUD */
#define PLPMTU_MAXSEGSNT 0x00000004 /* last segment sent was full seg */
bzUnsubmitted
Not Done
Inline Actions

#define<tab>FOO<tab>BAR<space>/* Blah. */ will work best.

bz: #define<tab>FOO<tab>BAR<space>/* Blah. */ will work best.
/* /*
* Structure to hold TCP options that are only used during segment * Structure to hold TCP options that are only used during segment
* processing (in tcp_input), but not held in the tcpcb. * processing (in tcp_input), but not held in the tcpcb.
* It's basically used to reduce the number of parameters * It's basically used to reduce the number of parameters
* to tcp_dooptions and tcp_addoptions. * to tcp_dooptions and tcp_addoptions.
* The binary order of the to_flags is relevant for packing of the * The binary order of the to_flags is relevant for packing of the
* options in tcp_addoptions. * options in tcp_addoptions.
*/ */
▲ Show 20 Lines • Show All 460 Lines • Show Last 20 Lines